Ni+(H2)n: Ligand bond energies for ground state ions

4 September 1998

Chemical Physics Letters 293 Ž1998. 503–510

Niqž H 2 / n: Ligand bond energies for ground state ions Paul R. Kemper, Patrick Weis, Michael T. Bowers

)

Department of Chemistry, The UniÕersity of California at Santa Barbara, Santa Barbara, CA 93106-9510, USA Received 12 May 1998; in final form 13 July 1998

Abstract The H 2 ligand binding energies in NiqŽH 2 . n were measured to be 17.3, 17.6, 11.3, 7.1 and 4.2 kcalrmol, for n s 1–5, respectively. The very weakly bound sixth ligand begins a new solvation sphere. Association entropies were also measured. MP2 calculations were done to determine geometries and vibration frequencies as well as the origin of the bonding. The observed changes in BDE with ligation are due to electronic rather than steric effects. Comparisons are made to the FeqŽH 2 . n , CoqŽH 2 . n , CuqŽH 2 . n and NiqŽCO. n systems. A highly symmetric D 3h planar structure is found in NiqŽH 2 . 3 and forms the core of the NiqŽH 2 .4, 5 ions. q 1998 Published by Elsevier Science B.V. All rights reserved.

1. Introduction Over the past ten years systematic experiments have examined a large number of gas-phase transition metal ion MqP X n clusters. Relatively simple systems with X s H 2 w1–9x, CO w10,11x, and the rare gases w12x, have been examined, as well as saturated and unsaturated hydrocarbons and other species Žfor a general discussion, see Ref. w13x.. The strong influence of the metal ion species on both bond energies and structures in these clusters clearly shows the presence of covalent forces in the bonding. Experimental and theoretical investigations have identified many of these. First, electron donation from the H 2 s orbital to the metal stabilizes the ion charge Žw14,15x; for a general discussion, see Ref. w16x.. Most of this donation is to the metal 4s orbital with a minor amount to the 3ds orbital. Second, metal ions with filled 3dp orbitals donate into the H 2 s ) )

Corresponding author. Fax: q1-805-893-4120.

orbitals w16,15,17,18x. This promotes s donation to the ion by returning electron density to the H 2 ligands, and increases the 3d–3d exchange stabilization on the metal. Third, in ions with half filled 3ds orbitals, the 3d z 2 and the 4s orbitals hybridize to reduce on-axis Pauli repulsion w16,15,17–19x. Fourth, the 4p orbitals, while significantly higher in energy, may play a significant role w16–18x. Finally, the non-covalent electrostatic interactions Žcharge-induced-dipole and charge quadrupole. are present, but usually comprise a small fraction of the total bond strength w1,16,19x. The relative importance of these five factors depends strongly on the valence configuration of the metal ion. The 4s orbital is especially critical and its occupation is always severely destabilizing w6,12,16x. In addition to experimental investigations a large number of theoretical calculations have been done on these MqP X systems. The Mq ŽH 2 . n clusters with M s Fe w7x, Co w15,17,18x and Cu w9x have been especially well characterized. These late transition

0009-2614r98r$ - see front matter q 1998 Published by Elsevier Science B.V. All rights reserved. PII: S 0 0 0 9 - 2 6 1 4 Ž 9 8 . 0 0 8 3 1 - 8

504

P.R. Kemper et al.r Chemical Physics Letters 293 (1998) 503–510

metal ions all have relatively strong bond dissociation energies ŽBDEs. and similar structures for the first and second clusters. Very different structures are found for larger numbers of ligands, however. In the third cluster, both Fe and Co assume a ‘T’ shape w7,17,18x while Cu has a planar D 3h geometry w9x. Both Fe and Co have six H 2 ligands in their first solvation spheres while Cu has four w7,9,17,18x. Nui et al. w20x have examined the NiqŽ H 2 .1, 2, 4, 6 ions using an unrestricted Hartree–Fock approach Žwith correlated valence electrons.. Their results for the first two clusters are in general agreement with our findings. Nui et al. predict, however, that Niq will add between six and ten H 2 ligands — a prediction at odds with previous theoretical and experimental studies of FeqŽ H 2 . n , CoqŽ H 2 . n and CuqŽ H 2 . n , and, as will be seen, with the present results. The geometries and bond energies for the larger clusters also disagree with the present results due to the assumption of D4h and O h symmetries for NiqŽ H 2 .4, 6 and possibly to the rather small basis sets used. In this Letter we present experimental association enthalpies and entropies for the sequential clustering of H 2 ligands to ground state Niq ions. These results are complimented by high-level ab initio calculations that give insight into the bonding mechanisms present. The NiqŽ H 2 . n clusters will be compared with other late transition metal clusters to illustrate the evolution of the bonding in the series. In addition, the NiqŽ H 2 . n clusters will be compared with previous results on the NiqŽ CO. n ions w11x to evaluate how well bonding concepts transfer from the H 2 to the CO systems.

2. Experimental Experimental details have been given previously w2,5,21x. Briefly, the Niq ions were formed either by glow discharge Žusing an Ar bath gas., by electron impact on NiŽCO.4 or by surface ionization of NiŽCO.4 . Electronic state chromatography experiments w22x showed that both ground state Ž2 D, 3d 9 . and electronically excited Niq) Ž2, 4 F, 4s1 3d 8 . were present when the discharge or electron impact sources were used Želectron impact ionization produced )

75% excited state w22x.. The excited states were resistant to collisional deactivation by H 2 and, because the 4s configuration binds only very weakly to H 2 w6x, caused a significant perturbation in the data for the first association. This was ultimately solved by using surface ionization which only forms ; 1% excited Niq. After formation, the isotope of interest Žeither 58 Niq or 60 Niq . is mass selected in the first quadrupole and injected into a driftrreaction cell filled with typically 10 Torr of H 2 . An equilibrium between the various NiqŽ H 2 . n cluster products Ž n s 0–6. is quickly established as the ions are moved through the 4 cm long cell with a small electric field. The field is small enough that no measurable perturbation of the thermodynamic temperature occurs. The ions then exit the cell and are mass analyzed in a second quadrupole. The resulting mass peaks are recorded and integrated and, together with the pressure of H 2 Ž p H 2 . and the temperature ŽT . are used to calculate the equilibrium constant Ž K p . and standard free energy Ž DG To . )

K po s

Niq Ž H 2 . n 760 )

Niq Ž H 2 . ny1 p H 2

DG To s yRT ln K po .

,

Ž 1. Ž 2.

Measurements were taken as a function of temperature from 77 to 780 K. The resulting plots of DG To vs. T were linear over the experimental temperature range, yielding D HTo and D STo as the intercept and slope, respectively. To determine the heat of reaction at 0 K Ž D H0o ., the experimental DG To vs. T curve is matched with a corresponding theoretical curve calculated using the theoretically determined structures and vibrational frequencies. The low vibrational frequencies and possible mass discrimination are varied to determine the value and uncertainty in the bond dissociation energy ŽBDE ' yD H0o s D 0 . w23x.

3. Theoretical methods The product ions discussed here were all examined theoretically both to determine the molecular parameters needed to analyze the experimental data and to identify factors important in the bonding.

P.R. Kemper et al.r Chemical Physics Letters 293 (1998) 503–510

Calculations were carried out at the MP2 level w24x using the TURBOMOLE package w25x. The hydrogen basis set was the ‘SVP’ set w26x. For the nickel basis set we used the ‘ECP-10-MDF’ which includes a pseudo-potential to describe the core electrons w27x.

4. Results and discussion Fig. 1 shows the experimental DG To vs. T data. The experimental enthalpies and entropies for the association reactions are summarized in Table 1 along with theoretical binding energies for both ground and excited state cluster ions. Calculated vibrational frequencies are listed in Table 2. Structures of the various NiqŽ H 2 . n ions are shown in Fig. 2. Finally, Table 3 compares BDEs for the first-row transition metal ion MqŽ H 2 . n complexes. 4.1. Ni q( H2 ) The NiqH 2 complex has an experimental BDE of 17.3 kcalrmol. This is similar to that found with the

505

3d n complexes of Feq ŽH 2 . and CoqŽ H 2 . Ž17.8 and 18.2 kcalrmol, respectively. but somewhat greater than that of CuqŽ H 2 . Ž15.4 kcalrmol, see Table 3.. Our calculated BDE is 15.6 kcalrmol, or ; 90% of the experimental value. The cluster structure is nearly identical to that of the other late metals: all have C 2v ˚ symmetry and an Mq–H 2 bond length of ; 1.65 A. We might expect that both the electrostatic and covalent attractions in CuqŽ H 2 . would be larger due to the slightly smaller ionic radius w28,29x; however, the NiqŽ H 2 . BDE is actually ; 15% greater. The origin of the lower bond strength in CuqŽ H 2 ., a 3d10 ion, is the filled 3d z 2 Žds . orbital on Cuq which destabilizes the complex. In both CoqŽ H 2 . and NiqŽ H 2 . the H 2 approaches a half-filled 3d z 2 orbital and calculations indicate 3dsr4s hybridization reduces the Mq–H 2 repulsion by moving the z axis electron density to the xy plane w16–19x. This hybridization is a higher-energy process in Cuq due to the necessity of moving two electrons and the resulting bond is weaker w9x. Binding energies were also calculated for the excited states of NiqŽ H 2 . which correspond to mov-

Fig. 1. Plot of experimental DGTo vs. temperature for the association reaction Niq ŽH 2 . ny1 q H 2 ™ NiqŽ H 2 . n . Open symbols refer to electron impact data Ž; 75% electronically excited Niq) ., closed symbols refer to surface ionization data Ž99% ground state Niq. . The excited Niq) only affects the data for the first association.

P.R. Kemper et al.r Chemical Physics Letters 293 (1998) 503–510

506

ing the electron hole in ground state Niq Ž2 D, 3d9 . to one of the other five 3d orbitals ŽTable 1.. When the 3d x y Ž3dd . is half filled, the BDE is reduced by ; 1.5 kcalrmol. As expected, this corresponds closely to the BDE for CuqŽ H 2 . since the Mq–H 2 interactions are now very similar. A major reduction in BDE Ž7.7 kcalrmol. occurs when the back bonding 3dp is half filled Žthe 2 B 1 state., showing the importance of this interaction. As expected, the BDE is now similar to the early metals ŽTable 3.. The NiqŽ CO. n ions have been investigated both experimentally and theoretically and exhibit bonding very similar to those in NiqŽ H 2 . n w11,30x. The carbon s orbital donates into the Niq 4s orbital while the filled Niq dp orbitals Ž3d x z and 3d y z . donate into the CO pp ) orbitals. Hybridization of the Niq 4s and 3ds orbitals is used to reduce Pauli repulsion. Again, the s and p dative interactions are synergistic: the p donation makes the CO more negative and thus better able to donate into the Mq 4s orbital. The strength of the Niq–CO bond is much greater than that of Niq H 2 Ž41.711 vs. 17.3 kcalrmol. at least partly due to the presence of two p interactions

instead of one. Both MqŽ CO. and MqŽ H 2 . show similar trends as the Mq ion is changed from Coq to Ni q to Cuq, but the Mq–H 2 bond strengths are ; 40–45% of the MqCO values. 4.2. Ni q( H2 )2 The second H 2 ligand binds opposite the first in D 2h symmetry with a measured BDE of 17.6 kcalrmol. The calculated bond lengths are nearly identical to those in NiqŽ H 2 ., reflecting the similarity in bonding ŽFig. 2.. The second BDE is 0.3 kcalrmol larger than that of the first cluster, however, due to the benefit of the symmetric 4sr3ds hybridization which occurred with the first H 2 addition w19x. The D 2d structure is calculated to be 0.8 kcalrmol higher in energy than the D 2h despite the fact that in D 2h symmetry back donation to both ligands is from a single p-type orbital Žd y z .. In both CoqŽ H 2 . 2 and CuqŽ H 2 . 2 the D 2h states are higher in energy Ž; 0.7 and ; 0.3 kcalrmol, respectively.. This reverse preference is also present in VqŽ H 2 . 2 Ž3d 4 , D 2h . and may be due to the availability of a

Table 1 Data summary for NiqŽ H 2 . n clusters Ion

Experiment D H0o a

Theory yD HTo a

yDSTo b

T

c

e

18.7 " 0.5

20.7 " 2

650 " 90

Niq ŽH 2 . 2

17.6 " 0.3 e

18.0 " 0.4

24.1 " 2

Niq ŽH 2 . 3

11.3 " 0.3 e

11.7 " 0.4

e

4.2 " 0.2 e

qŽ

Ni

H2.

qŽ

Ni

H 2 .4

Niq ŽH 2 .5 qŽ

Ni a

H 2 .6

17.3 " 0.3

7.1 " 0.3

0.8 " 0.2

g

2

De d

D0 d

C 2v A 1 ex. 2A 2 ex. 2 B1 ex. 2 B 2

17.9 16.5 10.2 13.1

15.6

575 " 175

D 2h D 2d

19.4 18.6

15.5

23.5 " 2

425 " 175

D 3h plan.f

14.1

9.6

8.7 " 0.6

24.5 " 2

300 " 120

C s pyr. D4h vert.f

10.7 4.8

7.1

4.9 " 0.4

22.5 " 2

180 " 70

Cs bipyd.

5.3

0.9 " 0.2

10.0 " 2

120 " 40

In kcal moly1 . In cal moly1 Ky1 . c In kelvin Ž" refers to temperature range, not uncertainty.. d MP2 geometries. e Fitting with theoretical frequencies and geometries. f The H–H bonds are all either perpendicular Žvert.. or parallel Žplan.. to the sh plane. g Fitting with DC p correction Žerror small at low temperature.. h Sixth H 2 is unbound in first solvation shell. b

symmetry d

D 2h

y8.5

3.2 e, h

–

b

All numbers in cmy1 . Frequencies in bold refer to the Niq ŽH 2 . 3 D 3h core. c Structure optimized to better than 1 mhartree; the negative frequency is associated with the very loose orientation of the fourth H 2 ligand.

507

a

874 1009, 899 983, 971(2) 964, 948, 945, 730 900, 889, 888, 660, 597 3835 3861, 3840 3604(2), 3603 4178, 3648, 3645, 3616 4261Ž2., 3666, 3664, 3654 4515 Niq ŽH 2 . Niq ŽH 2 . 2 Niq ŽH 2 . 3 Niq ŽH 2 .4 Niq ŽH 2 .5 H2

Asym. Mq –H 2 stretch H–H stretch Ion

Table 2 Theoretical vibrational frequencies for Niq ŽH 2 . n clustersa,b

The observed bond dissociation energy for the third H 2 ligand is 11.3 kcalrmol ŽTable 1.. This represents a decrease in average BDE from 17.4 to 15.4 kcalrmol. The optimized MP2 structure for the NiqŽ H 2 . 3 ion has D 3h symmetry with all atoms in the xy plane ŽFig. 2.. This same structure is found in CuqŽ H 2 . 3 and as discussed previously, both the highly symmetric structure and the marked decrease in bond strength are mainly due to the repulsive interaction of the H 2 s orbitals with the 3d shell on Cuq and Niq. In the planar D 3h structure, the three H 2 ligands have a repulsive s interaction with three o f th e N i q 3 d o rb ita ls Ž 3 d z 2 Ž a 1 . a n d 3d x 2yy 2r3d x y ŽeX ., the z axis is the symmetry axis.. The ion reduces this repulsion in two ways. First, the 4s and 3d z 2 orbitals hybridize to move electron density from the xy plane to the "z lobes. Second, and more important, the Niq ion donates electron density from the repulsive 3d x 2yy 2r3d x y ŽeX . pair of orbitals to the three H 2 s ) orbitals. This pair of orbitals is symmetry adapted to produce both two repulsive s interactions and two attractive p interactions with the three ligands. The resulting p back donation is similar to that in the first two clusters, but in the third cluster the back donation also reduces the s repulsion. This ‘push–pull’ mechanism has the effect of greatly increasing the extent of back donation while at the same time reducing s repulsion and is the primary driving force for the D 3h structure. The large increase in H–H bond length ŽFig. 2. and reduction in H 2 vibrational frequency ŽTable 2. show the presence of increased back donation in the third cluster. The second Žvertical. D 3h structure, with the H 2 ligands perpendicular to the xy plane, is higher in energy and is actually a transition state. This might seem surprising since the vertical and planar D 3h forms have nearly identical s repulsion and each has

Sym. Mq –H 2 stretch

4.3. Ni q( H2 )3

1378 1401, 1386 1450(2), 1444 1429, 1416, 1412, 1098 1405, 1401, 1400, 938, 937

H 2 –Mq –H 2 bends and rotations

second 2A g state to mix with the ground state in D 2h symmetry Žabsent in D 2d symmetry.. The NiqŽ CO. 2 ion is also linear but, unlike the in H 2 case, the second CO BDE is slightly weaker than the first Ž40.1 vs. 41.7 kcalrmol w11x.. In both the first and second clusters, however, the Niq–H 2 bond strength is 42 " 2% of the corresponding CO BDE.

434, 372, 189 635, 611Ž2., 429, 378Ž2. 590, 560, 550, 454, 443, 415, 398, 391, y68 c 570, 570, 558, 554, 486, 414, 413, 367, 351, 306, 82, 21

P.R. Kemper et al.r Chemical Physics Letters 293 (1998) 503–510

508

P.R. Kemper et al.r Chemical Physics Letters 293 (1998) 503–510

a degenerate pair of d orbitals available for back bonding Žeither the eY or eX set.. Further, the vertical isomer has lower ligand–ligand repulsion Žalthough this is negligible.. The destabilization of the vertical D 3h cluster is due to the loss of the very beneficial ‘push–pull’ mechanism present in the planar geometry since back donation from the eY pair of orbitals does not lead to any reduction in s repulsion. The relative stabilities of the planar D 3h geometries Žfound in NiqŽ H 2 . 3 and CuqŽ H 2 . 3 . and the asymmetric ‘T’ structures Žin FeqŽ H 2 . 3 and CoqŽ H 2 . 3 . have been discussed previously w9x.

A very similar reduction in BDE is found between the NiqŽ CO. 2 and NiqŽ CO. 3 w11x. Although no calculations have been done on NiqŽ CO. 3 , a D 3h ground state is found in the iso-electronic FeyŽ CO. 3 ion w31x. 4.4. Ni q( H2 )4 The fourth H 2 ‘caps’ the planar third cluster to form a distorted trigonal pyramid. The symmetry is C s , and Fig. 2 shows that the three equatorial ligands are strongly bound while the polar ligand is much more weakly attached. There is little change in the

Fig. 2. Theoretical geometries of the NiqŽ H 2 .1 – 5 ions calculated at the MP2 level. All distances are in angstrom. ˚ ¨

P.R. Kemper et al.r Chemical Physics Letters 293 (1998) 503–510 Table 3 Mq Ž3d n .: comparison of binding energiesa for Mq ŽH 2 . ny1 qH 2 ™ Mq ŽH 2 . n n 1 2 3 4 5 6

Tiq 10.0 9.7 9.3 8.5 8.2 8.7

Vq b

10.2 10.7 8.8 9.0 4.2 9.6 e

Crq 7.6 9.0 4.7 3.4 1.4 1.1

Feq 16.5 15.7 7.5 8.6 2.2 2.3

c

Coq

Niq

Cuq

18.2 17.0 9.6 9.6 4.3 4.0

17.3 17.6 11.3 7.1 4.2 0.8 d

15.4 16.7 8.8 5.1 1.0 d 1.0

a

In units of kcal moly1 . Binding energy with respect to lowest d 3 configuration. c Binding energy with respect to lowest d7 configuration. d Start of next solvation sphere. e The increase in y D H0o for ns6 is due to a spin change from quintet to triplet on the core Vq ion Žsee Refs. w4,16x.. b

Niq–H 2 or H–H bond lengths of the equatorial H 2 ligands between the third and fourth clusters and the very stable D 3h core is largely preserved. The experimental BDE is 7.1 kcalrmol, which is 8.3 kcalrmol less than the three equal 15.4 kcalrmol bonds in the third cluster. The decrease is due primarily to increased s repulsion from the 3d z 2 orbital. Remember that in the third cluster 4sr3d hybridization was used to reduce s repulsion by moving the 3d z 2 equatorial electron density to the "z poles. This same hybridization now works to greatly increase repulsion toward the fourth H 2 ligand and a lower BDE results. This structure is unique among the transition metal–H 2 complexes. The CuqŽ H 2 .4 ion has a quasi-tetrahedral geometry with four nearly equal bonds. The NiqŽ H 2 .4 geometry Žcapped D 3h . is unstable in Cuq due to the increased repulsion from the filled 3d z 2 orbital. Both FeqŽ H 2 .4 and CoqŽ H 2 .4 have ‘butterfly’ structures with two strongly bound ligands on the z axis and two more weakly bound on the x and y axes w7,17,18x. This is due to the larger number of half-filled orbitals which stabilize the side-on Ž‘T’. geometry relative to the planar D 3h in the third and fourth clusters. The NiqŽ H 2 . 3 D 3h geometry is actually unstable in FeqŽ H 2 . 3 and CoqŽ H 2 . 3 due to an unequal occupation of the eX orbital set which causes a Jahn–Teller distortion of the D 3h symmetry. These effects have been discussed in detail previously w9x. The BDE of NiqŽ CO.4 is also much lower than qŽ Ni CO. 3 w11x but no calculations on these ions have

509

been done. However, the isoelectronic FeyŽ CO.4 ion is formed by capping the planar FeyŽ CO. 3 ion w31x. 4.5. Ni q( H2 )5 Our MP2 calculations show the NiqŽ H 2 .4 ion ground state to be a trigonal bipyramid ŽFig. 2. with two weakly bound polar H 2 ligands capping the planar NiqŽ H 2 . 3 core group. Both the bond lengths and the H 2 vibrational frequencies ŽTable 2. show that the NiqŽ H 2 . 3 core is largely unchanged from the third cluster, indicating again that this structure is extremely stable. The BDE of the fifth H 2 is 4.2 kcalrmol, down form the 7.1 kcalrmol in the fourth cluster. This is partly due to destabilization of the NiqŽ H 2 . 3 core Žthe Niq–H 2 bond lengths increase ˚ . but it is the fourth H 2 very slightly, ; 0.01 A ligand which seems to be most affected Žhere the ˚ .. We speculate Niq H 2 bond length increases 0.13 A that the 4p z orbital is used in the fourth cluster to polarize the z axis electron density away from the polar H 2 ligand as in FeyŽ CO.4 w31x. This mechanism is counter-productive when the fifth H 2 ligand is added opposite the fourth, and a reduction in BDE occurs. The fifth addition closes the first solvation sphere for the NiqŽ H 2 . n clusters. Both Feq and Coq Žas well as all the early transition metal ions. add six H 2 ligands in their first solvation shell while Cuq adds four. The determining factor is the occupation of the Cartesian orbital set Žthe 3d x 2yy 2 and 3d z 2 orbitals. w9x. If these orbitals are each empty or half filled, octahedral occupation occurs; as the set is filled, s repulsion increases and the first shell occupation is reduced to five ŽNiqŽ H 2 .5 . or four ŽCuqŽ H 2 .4 .. 4.6. Ni q( H2 )6 Both the extremely weak BDE Ž0.8 kcalrmol. and, especially, the very small D STo show that the sixth ligand is added in the second solvation shell. The very low vibrational frequencies, free H 2 rotations and large rotational moments which result all give rise to more product entropy and make the D STo of association more positive Žy10.5 cal moly1 Ky1 vs. y24 to y21 cal moly1 Ky1 for the first five association reactions, see Table 1.. The lowest energy first solvation shell structure for NiqŽ H 2 . 6 is

510

P.R. Kemper et al.r Chemical Physics Letters 293 (1998) 503–510

calculated to be octahedral Žas in CoqŽ H 2 . 6 .; however, this structure is unbound with respect to the separated reactants ŽTable 1..

Acknowledgements The support of the National Science Foundation under grant CHE-9729146 is gratefully acknowledged. PW would like to thank the ‘‘Deutscher Akademischer Austauschdienst’’ for financial support.

References w1x J.E. Bushnell, P.R. Kemper, M.T. Bowers, J. Phys. Chem. 98 Ž1994. 2044. w2x J.E. Bushnell, P.R. Kemper, P. Maıtre, ˆ M.T. Bowers, J. Am. Chem. Soc. 116 Ž1994. 9710. w3x J.E. Bushnell, P.R. Kemper, P. Maıtre, M.T. Bowers, J. ˆ Chem. Phys. 106 Ž1997. 10153. w4x J.E. Bushnell, P.R. Kemper, M.T. Bowers, J. Phys. Chem. 97 Ž1993. 11628. w5x P.R. Kemper, P. Weis, M.T. Bowers, Int. J. Mass Spectrom. Ion Phys. 160 Ž1997. 17. w6x P. Weis, P.R. Kemper, M.T. Bowers, J. Phys. Chem. A 101 Ž1997. 2809. w7x J.E. Bushnell, P.R. Kemper, M.T. Bowers, J. Phys. Chem. 99 Ž1995. 15602. w8x P.R. Kemper, J. Bushnell, P.A.M. van Koppen, M.T. Bowers, J. Phys. Chem. 97 Ž1993. 1810. w9x P.R. Kemper, P. Weis, P. Maıtre, M.T. Bowers, J. Am. ˆ Chem. Soc. Žsubmitted.. w10x F. Meyer, Y.-M. Chen, P.B. Armentrout, J. Am. Chem. Soc. 117 Ž1995. 4071, and references therein. w11x F.A. Khan, D.L. Steele, P.B. Armentrout, J. Phys. Chem. 99 Ž1995. 7819.

w12x P.R. Kemper, M.-T. Hsu, M.T. Bowers, J. Phys. Chem. 95 Ž1991. 10600. w13x B.S. Freiser ŽEd.., Organometallic Ion Chemistry, Kluwer, Dordrecht, 1996. w14x P. Maıtre, C.W. Bauschlicher, J. Phys. Chem. 99 Ž1995. ˆ 6836. w15x P. Maıtre, C.W. Bauschlicher, J. Phys. Chem. 97 Ž1993. ˆ 11912. w16x C.W. Bauschlicher, H. Partridge, S.R. Langhoff, in: B.S. Freiser ŽEd.., Organometallic Ion Chemistry, Kluwer, Dordrecht, 1996. w17x P. Maıtre, C.W. Bauschlicher, J. Phys. Chem. 99 Ž1995. ˆ 3444. w18x J.K. Perry, G. Ohanessian, W.A. Goddard, J. Phys. Chem. 97 Ž1993. 5238. w19x H. Partridge, C.W. Bauschlicher, S.R. Langhoff, J. Phys. Chem. 96 Ž1992. 5350. w20x J. Niu, B.K. Rao, P. Jena, Phys. Rev. Lett. 68 Ž1992. 2277. w21x P.R. Kemper, M.T. Bowers, J. Am. Soc. Mass Spectrom. 1 Ž1990. 197. w22x M.T. Bowers, P.R. Kemper, G. von Helden, P.A.M. van Koppen, Science 260 Ž1993. 1446. w23x P.R. Kemper, M.-T. Hsu, M.T. Bowers, J. Phys. Chem. 95 Ž1991. 10600. w24x C. Møller, M.S. Plesset, Phys. Rev. 46 Ž1934. 618. w25x R. Ahlrichs, M. Bar, H. Horn, C. Kolmel, Chem. ¨ M. Haser, ¨ ¨ Phys. Lett. 162 Ž1989. 165. w26x A. Schafer, H. Horn, R. Ahlrichs, J. Chem. Phys. 97 Ž1992. ¨ 2571. w27x M. Dolg, U. Wedig, H. Stoll, H. Preuss, J. Chem. Phys. 86 Ž1987. 866. w28x L.A. Barnes, M. Rosi, C.W. Bauschlicher, J. Chem. Phys. 93 Ž1990. 609. w29x P. Desclaux, At. Data Nucl. Data Tables 12 Ž1973. 312. w30x L.A. Barnes, M. Rosi, C.W. Bauschlicher, J. Chem. Phys. 93 Ž1990. 609. w31x A. Ricca, C.W. Bauschlicher, J. Phys. Chem. 99 Ž1995. 5922.