J. Phys. Chem.

A 2001, 105, 4851-4864 4851

Singlet and Triplet Reaction Paths for Gas-Phase Zr + C2H4 by Density Functional Theory

Meredith Porembski and James C. Weisshaar*

Department of Chemistry, 1101 UniVersity AVenue, UniVersity of Wisconsin-Madison,
Madison, Wisconsin 53706-1396
ReceiVed: January 19, 2001; In Final Form: March 23, 2001

We use the B3LYP density functional theory with a large basis set to characterize stationary points on both
triplet and singlet potential energy surfaces for the gas-phase reaction Zr + C2H4 f ZrC2H2 + H2. The
previously described stepwise rearrangement path occurs on the triplet surface, requiring passage over a
substantial exit channel barrier. A new, lower energy triplet path involves concerted rearrangement of the
HZrC2H3 insertion intermediate directly to a weakly bound, product-like complex with no exit channel barrier
to triplet products. A new low-energy singlet path involves stepwise rearrangement from HZrC2H3 to the
strongly bound dihydrido species H2ZrC2H2, which then dissociates to singlet products over a small exit
channel barrier of 4 kcal/mol. We argue that the singlet path is more consistent than either triplet path with
the experimental product kinetic energy distribution, which peaks at 3-5 kcal/mol. This in turn suggests that
access of the singlet surface via fast intersystem crossing from the triplet to singlet metallacyclopropane
complex competes effectively, perhaps dominating at low collision energy. As in earlier work, B3LYP places
key transition state energies too high by 6-9 kcal/mol. The mPW1PW91 density functional gives much
more realistic energies.

I. Introduction the interaction of second-row transition metals with small

alkenes.13,16-18 The experimental and theoretical studies agree
Several recent studies have established the combination of that the dominant Zr + C2H4 reaction path involves bimolecular
experiment and electronic structure theory as a powerful elimination of H2 from the alkene, although detailed mechanistic
approach to elucidation of reaction mechanisms of bare, gas-
questions remain.
phase transition metal atoms with hydrocarbons.1-5 The well-
studied metal cation reactions of Fe+, Co+, and Ni+ with ethane, The presumption thus far has been that the Zr (4d25s2, 3F) +
propane, and n-butane1,2,5-8 have served as benchmarks to test ethylene reaction proceeds exclusively on the triplet PES.13
the ability of density functional theory to accurately describe Primary CH bond insertion is well understood, proceeding by
key transition state energies in such electronically complex the formation of a long-lived metallacyclopropane complex,
organometallic systems. The popular hybrid density functional followed by metal insertion into a CH bond of ethylene. Most
B3LYP9 has led to a comprehensive mechanistic picture of these recently, the absence of a kinetic isotope effect for the Zr +
CC and CH bond activation processes. Careful comparison to C2D4 reaction combined with statistical rate modeling revealed
all the available experimental data, aided by statistical rate that CH bond insertion by the Zr atom must be a facile step
modeling, suggests that B3LYP typically overestimates rate- that does not control the overall reaction efficiency.4 Instead,
determining transition states by 5-7 kcal/mol. Nonetheless, it the rate-determining step is initial formation of the strongly
is well established that B3LYP captures the essential features bound ZrC2H4 complex. This may be inefficient either due to a
of the potential energy surface (PES) that govern transition metal small approach barrier or nonadiabatic effects in the entrance
cation chemistry with much less effort than computationally channel. In contrast, subsequent steps in the mechanism of H2
demanding multireference configuration interaction approaches elimination from the CH insertion intermediate HZrC2H3 are
such as CCSD(T) theory.10,11 not well characterized. Theory indicates that a low-energy
By comparison, neutral transition metal atom chemistry has stepwise path exists, involving β-hydrogen transfer to form a
been less extensively explored, although the body of experi- strongly bound dihydrido intermediate H2ZrC2H2.13
mental and theoretical data is expanding, particularly for the To date, no study has explicitly confirmed the stepwise
early second-row transition metals Y, Zr, and Nb.12,13 We have mechanism, investigated the possible existence of an exit
studied the chemical kinetics of Zr (4d25s2, 3F) with ethylene channel barrier to H2 elimination, or considered the possible
and propylene at 300 K using a fast flow reactor equipped with role of the lowest singlet PES in the Zr + ethylene reaction.
detection by photoionization mass spectrometry (PIMS) at 157 This paper presents such a detailed theoretical study at the level
nm.14,4 Davis and co-workers have studied the chemical of B3LYP density functional theory using a large basis set,
dynamics of the Zr + C2H4 reaction, using a crossed-beam including complete characterization of all intermediates and
apparatus and PIMS detection at 157 nm at higher collision transition state structures along both the triplet and singlet
energies (6-23 kcal/mol).15 In the past decade, Siegbahn, potential energy surfaces. Because many studies have found that
Blomberg, and others have established a theoretical picture of B3LYP overestimates key barrier heights, we compare the
ability of two other hybrid DFT methods, mPW1PW9119,20 and
* Author to whom correspondence should be addressed. E-mail: B1LYP,19 to provide accurate stationary point energies. We find
[email protected]. that the triplet PES branches from the insertion intermediate
10.1021/jp010219f CCC: $20.00 © 2001 American Chemical Society
Published on Web 05/02/2001
4852 J. Phys. Chem. A, Vol. 105, No. 20, 2001 Porembski and Weisshaar

TABLE 1: Calculated Energies (kcal/mol) of Stationary Points along the Triplet PESa
UB3LYP/ UmPW1PW91/ UB1LYP/
MCPF+ Stuttgart+ Stuttgart+ Stuttgart+
species PCI-80b 6-311++G(d,p) 6-311++G(d,p)c 6-311++G(d,p)c
3TS
ent +0.78 (+0.68) -2.2 +1.2
Zr-C2H4 (3B1) (1a) -9.7 (-9.9)
ZrC2H4 (3A2) (1b) -38.0 -28.2 (-28.0) -38.9 -25.0
3
TSins +1.8 +1.9 (+3.1) -6.2 +5.6
nonplanar HZrC2H3 -28.7 -27.4 (-26.3) -33.1 -25.2
planar HZrC2H3 (2b) -31.6 (-30.4) -38.3 -29.2
3
TSβH -27.5d -9.8 (-7.5) -16.0 -6.4
3
MCTS -13.7 (-12.1) -24.4 -9.5
H2Zr-C2H2 (3a) -64.0d -18.2 (-15.8) -22.2 -15.5
H2-ZrC2H2 (3b) -15.7 (-14.3) -24.6 -11.9
3TS
exit +3.0 (+5.6) -2.7 +7.2
Zr-C2H2 (3B1) + H2 (4a) +22.0 (+24.4)
ZrC2H2 (3A2) + H2 (4b) -18.5 -14.9 (-12.7) -21.8 -11.6
ZrH2 (3B1) + C2H2 +18.7e +13.0 (+15.7)
a
Energies relative to free reactants, corrected for differential zero-point energy effects; parentheses denote energetics for Zr + C2D4. b MCPF+PCI-
80 calculations by Blomberg and Siegbahn from ref 13, unless otherwise indicated. c Single-point energies for all structures optimized at the level
of UB3LYP/Stuttgart 6-311++G(d,p). d Although the geometry and spin state of these species were not described in ref 13, the energies most
likely refer to intermediates along the singlet spin surface (compare to 1TSβH and 7 in Table 3; see text for details). e Estimated exothermicity using
the heat of formation of C2H2 and the calculated binding energy of ZrH2 (3B1) from Siegbahn, P. E. M. Theor. Chim. Acta 1994, 87, 441.

HZrC2H3 into two low-lying paths. The higher energy, stepwise the Zr + ethylene PES, optimizing geometries and computing
rearrangement path passes through a dihydrido intermediate and energies using B3LYP/LANL2DZ (Los Alamos effective core
over a substantial exit barrier 3TSexit, lying 18 kcal/mol above potential (ECP) + DZ valence set for Zr and full DZ set for C
triplet products. The lower energy path directly connects and H). When possible, species calculated at the current level
HZrC2H3 to a product-like complex H2-ZrC2H2 via a multi- of theory will be explicitly compared to structures from these
center transition state 3MCTS. In this concerted rearrangement earlier studies.
mechanism, the product-like complex smoothly dissociates to For geometry optimizations of stationary points, potential
ZrC2H2 (3A2) + H2 products with little or no barrier. energy surface scans, and intrinsic reaction coordinate (IRC)
On the singlet PES, all transition states associated with searches from computed transition states, we combine the
primary CH insertion and stepwise H2 elimination are lower B3LYP density functional9 with the Stuttgart ECP + valence
than the analogous triplet transition states due to the greater set21 (triple-ζ quality) for Zr and the 6-311++G(d,p) basis set
bonding capacity of singlet-coupled Zr. We find strong evidence for C and H. We call this hybrid set Stuttgart+6-311++G(d,p)
of a low-energy, stepwise path from the singlet metallacyclo- and carry out the electronic structure calculations using the
propane complex to the exothermic ZrC2H2 (1A1) + H2 products GAUSSIAN-98 (G98) program, either at the spin-restricted (R)
over a small 4 kcal/mol exit channel barrier 1TSexit. The singlet or spin-unrestricted (U) level (described further below).22 The
path thus stands in significantly better qualitative agreement than basis set used in this study, while considerably larger than in
either of the low-energy triplet paths with the peaking of the our previous work (LANL2DZ), does not employ f functions
experimental product kinetic energy distribution at 3-5 kcal/ on the metal center. However, for this system, the structural
mol.15 Accordingly, we argue that intersystem crossing from and energetic parameters predicted by both LANL2DZ and
the triplet to singlet metallacyclopropane is likely to be Stuttgart+6-311++G(d,p) are in good agreement with the
important, especially at low collision energy and for small calculations using polarized functions at a higher level of
impact parameters. Finally, it appears that the mPW1PW91 theory.13 In addition, many reviews have noted that DFT
density functional provides much more accurate transition state methods are not overly sensitive to the size and quality of basis
energies in these systems than the B3LYP functional. sets.10,11
Throughout this paper, E denotes total energy in the reaction After optimization at the level of UB3LYP/Stuttgart+6-
complex, Et denotes translational energy of reactants, and Et′ 311++G(d,p), we compute single-point energies for most triplet
denotes translational energy of ZrC2H2 + H2 products. Unless stationary points using two other DFT methods, UmPW1PW91
otherwise indicated, all energies of stationary points are reported and UB1LYP. Single-point energies for RB3LYP optimized
relative to ground-state Zr (4d25s2, 3F) + C2H4 reactants, singlet stationary points were also obtained using RmPW1PW91.
including corrections for zero-point energy (ZPE) effects. As a test case, we reoptimized the ZrC2H4 (3A2) complex at
II. Methods the level of UmPW1PW91/Stuttgart+6-311++G(d,p), but found
The present work compares most directly to a series of that the energy decreased by only 0.1 kcal/mol; the geometry
theoretical studies by Siegbahn and others.16-18 They describe and vibrational frequencies remain essentially unchanged. Thus,
the geometries of potential minima and transition states for the energies calculated for nonoptimized structures with the different
Zr + C2H4 reaction optimized at the Hartree-Fock (HF) level functionals are likely to be realistic. This is consistent with the
and energies at the modified coupled pair functional (MCPF) experience that structural parameters predicted by B3LYP, even
level. Geometry optimizations used double-ζ (DZ) quality basis with modest basis sets, are of very good accuracy and, unlike
sets, while larger sets including diffuse and polarization func- energies, geometries and vibrational frequencies tend to be
tions on Zr were used for energy calculations. Later work13 insensitive to the choice of functional employed.10,11 In a few
repeats these calculations, but corrects the MCPF energies using cases on the triplet PES, a transition state found at the level of
the PCI-80 approximation (Table 1). In the latter study, structural UB3LYP/Stuttgart+6-311++G(d,p) actually disappears using
parameters and the spin identity of reaction intermediates were the other DFT methods and energies are calculated to lie below
not described. The most recent effort4 revisits some features of the reactant ground-state asymptote, resulting in reported “barrier
Reaction Paths for Gas-Phase Zr + C2H4 J. Phys. Chem. A, Vol. 105, No. 20, 2001 4853

heights” below zero. Nonetheless, we report such energies exhibit severe spin contamination, the spin-restricted calculations
because they are characteristic of the bond-breaking, bond- presumably provide an upper bound to the true singlet energy
forming processes with structures the same as the saddle points and have the additional property of being eigenfunctions of Ŝ2.
found with UB3LYP/Stuttgart+6-311++G(d,p). Deviations from a pure singlet or triplet description indicate
Electronic stability tests on all optimized stationary points that the true wave function would be better represented by a
were performed to ensure that the lowest energy solution to more sophisticated multi-configurational scheme, beyond the
the SCF equations was in fact found; all stationary points were scope of this study. Restricted-open shell DFT methods for
found to be stable with the exception of one (described below). singlets are currently unavailable within the G98 package.
For all transition states, an IRC analysis was performed,
following the path of steepest descent in both directions until III. Results
the geometries unambiguously approached those of the expected
A. Triplet PES. With the electronic wave function con-
minima.
strained to triplet spin multiplicity, we calculate those features
For a thorough treatment of the conceptual framework and of the Zr + ethylene PES illustrated in Figure 1. Below the
the practical and theoretical limitations of approximate density energy level diagram, we include the key structures that illustrate
functional theory, we refer the reader to excellent discussions the major atomic motions along the computed reaction paths,
by Koch et al.23 and other reviews of the method.24,25 However, while the detailed structural parameters of all reaction inter-
we do mention one important point here. Because it is the mediates are given in Figure 2. Table 1 summarizes the
N-electron ground-state density and not the N-electron waVe energetics of these species, using the Stuttgart+6-311++G-
function that is defined in DFT, and because the exact exhange- (d,p) basis set with three different density functional methods.
correlation functional is unknown, it remains unclear how to Also in Table 1, we collect results from prior theoretical studies.
deal with states that must be simultaneous eigenfunctions of The individual s and d gross populations and the natural charge
L̂2, L̂z, Ŝ2, and Ŝz for the case of atoms, or of Ŝ2 and Ŝ2 for the of the Zr metal center, as determined by a natural bond orbital
case of nonlinear polyatomic molecules. The pragmatic approach (NBO) analysis,28 are given in Table 2. Finally, rotational
has been to extract the corresponding quantum numbers from a constants and vibrational frequencies of all stationary points
single Slater determinant built from the spin-restricted Kohn- calculated at the level UB3LYP/Stuttgart+6-311++G(d,p) are
Sham (KS) orbitals themselves, but this is not always possible. collected in the Supporting Information to this paper.
Furthermore, the procedure fails if a state is not well-described In our earlier study,4 we searched for a possible barrier to
by a single Slater determinant. addition of ground-state Zr (4d25s2, 3F) to the alkene double
In particular, the magnitude of reported energy barriers and bond by scanning the distance R between Zr and the midpoint
binding energies depends on the choice of the reference atomic of the CC bond. The UB3LYP/LANL2DZ calculations found
ground state. Open-shell, spin-unrestricted atomic states that neither an approach barrier nor a barrier in the region of the
correspond to a single Slater determinant with a definite value curve-crossing (about 2.4 Å) between the Zr-C2H4 (3B1) long-
of L have been defined by electronic configurations with integer range complex (1a) and the ZrC2H4 (3A2) metallacyclopropane
occupancies of the real d orbitals as tabulated by Hay.26 complex (1b). These two rather different complexes are
Accordingly, the calculated reference atomic configuration that described below. At the level of UB3LYP/Stuttgart+ 6-311++G-
corresponds to the ground-state term Zr (4d25s2, 3F) is dz12 (d,p), we do find a small barrier of +0.78 kcal/mol to the
dx12-y2s2. Hay does not list the required occupation for the formation of the long-range complex.
excited-state term Zr (4d35s1, 3F), so we use the configuration The geometry of the entrance channel transition state 3TSent
d1.97 0.27 0.75 1
xz dz2 dx2-y2s to represent that atomic state. Unfortunately, is described in Figure 2. The transition state is located at R )
because we cannot apply symmetry constraints to atoms using 3.44 Å and there is minimal distortion of ethylene; compare
the current G98 package, unphysical mixing of the d orbitals structures 1a and 1b. The transition vector, also labeled, shows
can occur, leading to noninteger occupancies of d orbitals. the relative motion of Zr and ethylene along the reaction
Nevertheless, the calculated excitation energies of this con- coordinate, R. The calculated energy of 3TSent is +1.4 kcal/
figuration are 36, 34, and 37 kcal/mol using UB3LYP, mol, but the saddle point was determined to be electronically
UmPW1PW91, and UB1LYP, respectively. The experimental unstable. A lower energy solution finds that 3TSent decreases
excitation energy from ground-state Zr (4d25s2, 3F) to Zr (4d3- to +0.78 kcal/mol, as reported in Table 2. It is not expected
5s1, 3F) is 33 kcal/mol.27 The agreement between experiment that the structural parameters or vibrational frequencies of 3TSent
and theory suggests that the energies of chemically bound will change significantly in a reoptimization starting from the
species will be reasonable. Because the lowest energy singlet stabilized electronic state, and none was performed. The basis
atomic state cannot be represented by a single Slater determi- set superposition error (BSSE)29 at R ) 3.5 Å was previously
nant, we do not compute the excitation energy to the excited- found to be quite small, +0.04 kcal/mol. The BSSE at R )
state Zr (4d25s2, 1D); it is 14 kcal/mol experimentally.27 3.44 Å, the location of 3TSent, should be equally small.4 Finally,
Finally, we note that for open-shell species, the value of 〈Ŝ2〉 although the PES in the entrance channel is obviously very flat,
3TS
can be calculated using the determinant built from the spin- ent does provide a set of vibrational frequencies and rotational
unrestricted KS orbitals and compared to the expected magnitude constants for the RRKM model (Section IV-B).
for the true triplet or singlet wave function.23 When the spin Thus the initial interaction of the metal and alkene leads to
contamination of the unrestricted determinants is significant, the formation of a covalently bound complex over at most a
the validity of computed structural and energetic parameters is small energy barrier. We find that Zr binds to ethylene in two
viewed with suspicion. Here, all open-shell triplet calculations distinct modes with triplet spin multiplicity. The geometry and
were performed at the spin-unrestricted level and deviations of energy of Zr-C2H4 (3B1) (1a) are described in Figure 2 and
the 〈Ŝ2〉 expectation values were less than 1% for all optimized Table 1. Because R is longer, we designate 1a the long-range
structures except two, as noted below. Generally, singlet complex, rather than a true metallacyclopropane species. The
calculations were performed at both the restricted and unre- distortion of ethylene is minimal, as indicated by only a slight
stricted level. In the cases where the unrestricted determinants puckering of the CH bonds and a short CC bond (1.38 Å). The
4854 J. Phys. Chem. A, Vol. 105, No. 20, 2001 Porembski and Weisshaar

Figure 1. UB3LYP/Stuttgart+6-311++G(d,p) reaction path energetics along the triplet PES. All energies measured relative to ground-state Zr
(4d25s2, 3F) + C2H4 reactants and corrected for zero-point energy.

geometry and energy of ZrC2H4 (3A2) (1b) are also included in charge of +0.8). Other calculations have also found the dn-1s1
Figure 2 and Table 1. The geometry of this species is nearly configuration to be the primary bonding state for early second-
identical to the one calculated at the level of UB3LYP/ row transition metals.16-18
LANL2DZ and to the metallacyclopropane complex calculated The bonding mechanism in the two structurally different
at the HF level.4,16 The binding energy of 1a is only one-third complexes follows the discussion outlined by Blomberg et al.,16
that of the strongly bound metallacyclopropane and the com- but will be repeated here for clarity. Essentially, to form the
puted value of 〈Ŝ2〉 shows a deviation of 9% from a pure triplet two metal-carbon σ bonds of either complex, the dyz, dz2, and
spin description. Although a similar long-range complex has dx2-y2 metal orbitals (with admixture of only about 12% of the
been described for Y-C2H4, the structure and energy of a long- metal s orbital) combine to maximize overlap with the pz orbitals
range, triplet Zr-C2H4 complex has not been explicitly re- on ethylene. There are six valence electrons to distribute among
ported.16 the ethylene π bond, two Zr-C bonding orbitals, and two metal
The electron configuration (d2.04s1.62) and natural charge nonbonding orbitals. In the long-range complex 1a, the Zr-C
(+0.32)28 of Zr in the long-range complex 1a reveals that this bond orbitals are singly occupied and one electron remains in
species retains much of the s2 character of the ground-state metal the ethylene π bond; in the metallacyclopropane 1b, both Zr-C
atom. In contrast, the metallacyclopropane 1b exhibits substan- bond orbitals are doubly occupied and the π bond in ethylene
tial contribution from an excited-state of the metal and a large is effectively broken. The nonbonding orbitals are another sd
shift of electron density to the hydrocarbon (d2.31s0.89 and natural hybrid orbital (∼75% s-character) and the dxy orbital, respec-
Reaction Paths for Gas-Phase Zr + C2H4 J. Phys. Chem. A, Vol. 105, No. 20, 2001 4855

calculated at the level of UB3LYP/LANL2DZ.4 Reference 13

reports the energy of a “new and lower” CH insertion transition
state than the one previously found (ref 17), but does not
describe its geometry. The transition vector for the imaginary
frequency, also labeled in Figure 2, clearly indicates the breaking
of the CH bond.
The insertion intermediate HZrC2H3 (2b) is described in
Figure 2 and Table 1. Most notably, 2b is planar. Other structural
parameters, including the small C2-C1-Zr bond angle, rela-
tively close contact to H2 and the elongation of the β-CH bond
are reminiscent of the stabilizing agostic interactions found in
many transition metal cation-hydrocarbon complexes.1 A non-
planar insertion intermediate (dihedral angle, ΘH3-C1-Zr-H4 )
128°) was also identified, with structural parameters more
similar to the geometry of HZrC2H3 reported in refs 17 and 4.
Its energy is also included in Table 2. Although all DFT methods
find the planar HZrC2H3 (2b) to be the global minimum, a
potential energy surface scan of the H3-C1-Zr-H4 dihedral
angle revealed that a negligible barrier (∼1 kcal/mol) separates
the two intermediates. Under all reaction conditions, with some
30 kcal/mol of internal energy, rapid conversion between planar
and nonplanar HZrC2H3 will occur. Nonetheless, at the current
level of theory, we find 2b to be the stationary point that is
unambiguously connected to 3TSins from reactants and to 3TSβH
and 3MCTS toward products.
After primary CH bond insertion, elimination of H2 from the
insertion intermediate requires transfer of the alkene βH to the
Figure 2. Details of UB3LYP/Stuttgart+6-311++G(d,p) optimized metal center. At this level of theory, we find that the triplet
geometries of potential minima and transition states for Zr + C2H4 PES branches from the planar insertion intermediate 2b. The
along the triplet PES. Distances in angstroms, angles in degrees. For large dashed path in Figure 1 follows the stepwise elimination
each transition state, the transition vector corresponding to the imaginary of H2 via 3TSβH, while the small dashed path follows the
frequency is shown. concerted elimination of H2 via 3MCTS. The latter is found to
TABLE 2: Natural Populations of Zr for Triplet Stationary be the lowest energy path on the triplet PES to exothermic
Points Using UB3LYP/Stuttgart+6-311++G(d,p) Theory ZrC2H2 (3A2) + H2 products.
natural The βH transfer transition state 3TSβH that carries the planar
species chargea 4d 5s insertion intermediate (2b) to a dihydrido intermediate (3a) is
3TS
ent
b +0.01 2.03 1.96 described in Figure 2 and Table 1. The labeled transition vector
Zr-C2H4 (3B1) (1a) +0.32 2.04 1.62 clearly corresponds to the expected β-CH bond stretch. The
ZrC2H4 (3A2) (1b) +0.80 2.31 0.89 resulting dihydrido intermediate H2Zr-C2H2 (3a), distinguished
3
TSins +0.62 2.44 0.93 by the formation of strong bonds to hydrogen and weak bonds
nonplanar HZrC2H3 +1.01 2.15 0.81 to acetylene, is described in Figure 2 and Table 1. The geometry
planar HZrC2H3 (2b) +1.01 2.21 0.76
3TS
+1.15 2.20 0.62 and bonding mechanism of the ZrC2H2 unit of the complex are
βH
3MCTS +0.71 2.50 0.79 virtually identical to the long-range Zr-C2H2 (3B1) complex
H2Zr-C2H2 (3a) +1.33 2.07 0.59 (4a) described below. Furthermore, the geometry of the ZrH2
H2-ZrC2H2 (3b) +0.69 2.51 0.80 unit of the complex is very similar to that calculated for ZrH2
3TS
exit +1.02 2.31 0.63 (3B1) products (ZrH ) 1.87 Å and HZrH ) 119°; structure not
Zr-C2H2 (3B1) (4a) +0.38 1.57 2.03
ZrC2H2 (3A2) (4b) +0.81 2.30 0.91
shown).
a The natural charge on Zr, in units of electrons, as found from NBO
However, we find a lower multi-center transition state
3MCTS (Table 1) that carries the insertion intermediate (2b)
population analysis; in all cases, the 5p population was less than 0.05.
b Values listed are for the electronically stable species; see text for directly to a product-like complex (3b). Figure 2 shows that
3MCTS is planar and the structural parameters indicate that the
details.
transition state occurs “late” on the potential energy surface.
tively. Both reduce repulsion with the alkene. In 1a, the sd The geometry and bonding mechanism of the ZrC2H2 unit are
orbital is doubly occupied. In 1b, one electron occupies the sd similar to the strongly bound ZrC2H2 (3A2) metallacyclopropene
orbital, while the second occupies the dxy orbital. Finally, in (4b) described below. Compared to 3TSβH, the ZrH distances
the long-range complex, the “half-π” bond donates some are slightly longer, 1.96 and 1.89 Å, and the H4-H2 distance
electron density (0.08 e-) to an empty sdz2 hybrid acceptor is much smaller, 0.99 Å. The NBO analysis shows donation of
orbital. σHH electron density (0.22 e-) into an empty sdx2-y2 hybrid
An essential step toward the elimination of H2 is insertion orbital on the metal, accompanied by back-donation (0.32 e-)
into a CH bond of ethylene by the metallacyclopropane complex. from the Zr-C2 bond to σHH*. The substantial donor-acceptor
The geometry and energy of the triplet CH insertion transition interactions among four centers, the two hydrogens, Zr, and C1,
state 3TSins that connects the metallacyclopropane (1b) to an inspire us to label this saddle point a multi-center transition state.
insertion intermediate (2b) are described in Figure 2 and Table The transition vector depicted in Figure 2 involving the
2. The nonplanar structure of 3TSins is very similar to the one simultaneous β-CH stretch and formation of molecular H2,
4856 J. Phys. Chem. A, Vol. 105, No. 20, 2001 Porembski and Weisshaar

further reveals the concerted rearrangement of atoms that elimination reaction path or the concerted H2 elimination
connects the insertion intermediate directly to the precursor reaction path along the triplet PES.
complex for H2 loss (3b). Finally, we note that the geometry at The geometry and energy of the transition state 3TSexit that
the terminal point of the IRC scan stepping from 3MCTS toward connects the dihydrido intermediate H2Zr-C2H2 (3a) to products
2b is planar, except that the Zr-H4 bond is flipped (ΘH3C1ZrH4 are given in Figure 2 and Table 1. To reach the ZrC2H2 (3A2)
) 180°). However, as discussed above, rotation of the Zr-H4 + H2 exit channel, 3a must make a significant geometry change;
bond is energetically facile. the transition vectors in Figure 2 involve the H4-Zr-H2 angle
The resulting H2-ZrC2H2 complex (3b) is best described as bend and reorientation of the acetylene group. The IRC analysis
a complex between the metallocyclopropene ZrC2H2 and determined that 3TSexit leads directly to another H2-ZrC2H2
molecular H2 (Figure 2 and Table 1). Compared to the dihydrido exit channel complex. This species, whose structure is shown
intermediate 3a, planar 3b is very product-like. Again, the in Figure 1 (3b′), appears to be an isomer of 3b. The donor-
ZrC2H2 unit of the complex is virtually identical to the ZrC2H2 acceptor interactions, σHH f dxz and dxy f σHH*, account for
(3A2) complex (4b) described below. The ZrH distances are long the perpendicular, rather than coplanar, arrangement of H2.
(2.16 and 2.08 Å) and the H4-H2 distance (0.79 Å) is nearly Complex 3b′ is nearly isoenergetic to the ZrC2H2 (3A2) + H2
that of the calculated equilibrium bond length of molecular exit channel and, based on the H2 dissociation scan from 3b
hydrogen, 0.744 Å. The NBO analysis shows that similar (discussed below), we do not expect any further barrier beyond
donor-acceptor interactions stabilize this complex, as described this complex. Thus, we consider 3TSexit to be the effective barrier
above for 3MCTS. Finally, although 3a and 3b differ greatly (∼18 kcal/mol) for formation of the dihydrido intermediate H2-
in geometry, their binding energies are nearly the same. Zr-C2H2 (3a) from isolated ZrC2H2 + H2 ground-state reac-
Furthermore, it is not surprising that both 3MCTS and H2- tants.
ZrC2H2 (3b), with similar product-like geometries and bonding In contrast, we find no exit channel barrier for the concerted
mechanisms, lie nearly isoenergetic to the ZrC2H2 (3A2) + H2 elimination of H2 along the triplet spin surface. Starting from
exit channel. the H2-ZrC2H2 (3b) minimum, we compute the energy while
Before discussing the triplet exit channel PES along the two scanning the distance between Zr and H4, optimizing the
reaction paths, stepwise H2 elimination and concerted H2 geometry at each fixed value of Zr-H4. At the level of
elimination, we describe the geometry and energetics of the UB3LYP/Stuttgart+6-311++G(d,p), we find that 3b dissociates
ZrC2H2 + H2 products themselves. As for ethylene, we find to ZrC2H2 (3A2) + H2 products with monotonically increasing
that Zr binds to acetylene in two distinct modes. The geometry energy.
and energy of the Zr-C2H2 (3B1) complex (4a) are described Finally, we note that the three DFT methods predict a wide
in Figure 2 and Table 1. Like the Zr-C2H4 (3B1) complex (1a), range of energies for all species along the triplet PES, as shown
4a binds at a long range; the ZrC bonds are longer, 2.30 Å, in Table 1. Compared to the B3LYP calculations, mPW1PW91
and the CC bond length of 1.26 Å is close to the calculated increases binding energies and lowers transition states relative
triple bond length in acetylene, 1.20 Å. No similar long-range to ground-state reactants by about 7 kcal/mol on average, while
Zr-C2H2 complex has been reported in the literature. The B1LYP decreases binding energies and increases transition states
geometry and energy of the ZrC2H2 (3A2) complex are also by about 3 kcal/mol on average. We assess the relative
described in Figure 2 (4b) and Table 1. We call this complex performance of the different theories in explaining the behavior
a metallacyclopropene based on its short ZrC bonds (2.09 Å) of the Zr + ethylene system below.
and long CC bond (1.35 Å); the computed CC bond length in B. Singlet PES. Analogous calculations for the Zr + ethylene
free ethylene is 1.33 Å. The geometry of 4b is similar to the system constrained to singlet spin multiplicity were carried out
structure reported earlier,18 optimized at the HF level. We at the spin-restricted level using the B3LYP density functional
calculate the binding energy of 4a to be only 18.3 kcal/mol, (RB3LYP/Stuttgart+6-311++G(d,p)). The results are displayed
and as for the long-range complex 1a, there is some spin in Figures 3 and 4 and Tables 3 and 4. In all cases, reoptimi-
contamination (12%) of the UB3LYP determinant. We compute zation at the spin-unrestricted level (UB3LYP) did not signifi-
a binding energy of 55.1 kcal/mol for 4b, in good agreement cantly change geometries or vibrational frequencies. For the
with the binding energy of 56.9 kcal/mol calculated at the MCPF closed-shell species 1TSins, HZrC2H3 (6), 1TSβH, and H2ZrC2H2
level.18 (7), the RB3LYP and UB3LYP energies were virtually identical.
The bonding mechanisms in the acetylene complexes parallel For the species with nonbonding, spin-paired electrons, including
those already described for the ethylene complexes. However, ZrC2H4 (1A1) (5), 1TSexit, and ZrC2H2 (1A1) (8), and for some
as previously noted by Siegbahn, the interaction of the out-of- points along the exit channel PES scan, the UB3LYP calcula-
plane π orbitals with the metal d orbitalssnot present in the tions significantly lowered energies relative to the RB3LYP
metal-ethylene complexessaccounts for the very strong bind- values. Substantial spin contamination of the UB3LYP deter-
ing energies of the early second row transition metals, Y, Zr, minants was observed for such structures, as reported in Table
and Nb with acetylene.18 Essentially, the out-of-plane acetylene 3. In these cases, we view the spin-restricted energies as likely
π orbital donates electron density into the unoccupied metal upper bounds to the true singlet surface, as described above.
dxz orbital accompanied by back-donation of dxy electron density Similar to the triplet PES, the RmPW1PW91 calculations yield
to the π* orbital. significantly lower energies compared to the RB3LYP values
The exit channel PES for the Zr + ethylene reaction governs for most singlet stationary points (Table 3). Finally, rotational
the disposal of energy into H2 elimination products. Thus far, constants and vibrational frequencies of all singlet stationary
we have shown the existence of a dihydrido intermediate H2- points calculated at the level of RB3LYP/Stuttgart+6-311++G-
Zr-C2H2 (3a) and also a bound complex between ZrC2H2 and (d,p) are collected in the Supporting Information to this issue.
molecular H2 (3b), whose structure is much more like ground- The singlet mechanism is qualitatively similar to the stepwise
state products. The corresponding rearrangement transition states triplet mechanism described above. The geometry and energy
are 3TSβH and 3MCTS. Next, we directly explore the connection of the ZrC2H4 (1A1) metallacyclopropane complex (5) are given
of 3a and 3b to products following either the stepwise H2 in Figure 4 and Table 3. The electron configuration of Zr
Reaction Paths for Gas-Phase Zr + C2H4 J. Phys. Chem. A, Vol. 105, No. 20, 2001 4857

Figure 3. RB3LYP/Stuttgart+6-311++G(d,p) reaction path energetics along the singlet PES. All energies measured relative to ground-state Zr
(4d25s2, 3F) + C2H4 reactants and corrected for zero-point energy.

(d1.55s1.74) indicates a correlation to the 4d25s2, 1D electronic and the CC bond length (1.48 Å) is close to that of free ethane,
state of the metal (Table 4). The bonding mechanism for 5 1.54 Å. As a result, 1TSins already has considerable metallacy-
follows the description of the ZrC2H4 (3A2) complex (1b) clopropane character; compare structures 5 and 6. Accordingly,
outlined above except that the two nonbonding electrons are at the level of RB3LYP/Stuttgart+6-311++G(d,p), 1TSins lies
now spin-paired in an sd hybrid orbital that reduces repulsion nearly 10 kcal/mol below 3TSins. The transition vector indicates
with ethylene. Not surprisingly, the geometries of the triplet the expected CH bond stretch and reorientation of the C1-H3
and singlet ZrC2H4 complexes are virtually identical. bond.
The geometry and energy of the CH insertion transition state The geometry and energy of the singlet insertion intermediate
1TS HZrC2H3 (6) are displayed in Figure 4 and Table 3. As in 1TSins,
ins that connects the metallacyclopropane (5) to an insertion
intermediate (6) is described in Figure 4 and Table 3. The Zr uses all four electrons to form a double bond to C1 and single
structure of 1TSins is quite different from 3TSins (Figure 2). The bonds to C2 and H4; the structure bears strong similarity to the
singlet transition state is “late,” with the CH bond fully broken metallacyclopropane complex (5).
and the ZrH bond essentially fully formed. In addition, Zr can In contrast to the triplet PES, we find a single low-energy
use all four electrons to form strong bonds to both carbons on path to ZrC2H2 (1A1) + H2 products involving the stepwise
ethylene and the hydrogen, as indicated by the short Zr-C1, elimination of H2 from the insertion intermediate (6). The βH
Zr-C2, and Zr-H4 distances. In fact, ZrC1 is a double bond transfer transition state 1TSβH that connects 6 to a strongly bound
4858 J. Phys. Chem. A, Vol. 105, No. 20, 2001 Porembski and Weisshaar

TABLE 4: Natural Populations of Zr for Singlet Stationary

Points Using RB3LYP/Stuttgart+6-311++G(d,p) Theory
natural
species chargea 4d 5s
1
ZrC2H4 ( A1) (5) +0.72 1.55 1.74
1TS
ins +0.82 1.09 2.11
HZrC2H3 (6) +1.32 0.44 2.28
1
TSβH +1.19 2.46 0.37
H2ZrC2H2 (7) +1.64 1.93 0.45
1TS
exit +0.69 1.81 1.50
ZrC2H2 (1A1) (8) +0.71 1.59 1.71
a
The natural charge on Zr in units of electrons, as found from NBO
population analysis; in all cases, the 5p population was less than
0.01.

primarily this β-CH bond stretch and also the reorientation of

the acetylene group. Neither the structural parameters nor the
identity of the spin state of the earlier βΗ transfer transition
state were reported,13 but its energy is given in Table 1.
Comparing the energies of 3TSβH and 1TSβH (Tables 1 and 3)
to the PCI-80 calculation suggests that the original transition
state was a singlet species.
The geometry and energy of the H2ZrC2H2 complex (7) are
shown in Figure 4 and Table 3. The singlet dihydrido intermedi-
ate (7) is more strongly bound than the triplet dihydrido
intermediate (3a) by over 30 kcal/mol. In contrast to its triplet
counterpart, the closed-shell complex can form strong bonds
with both H2 and C2H2, as revealed by the short ZrH and ZrC
bond lengths. The geometry of 7 is distinctly asymmetric, with
both hydrogens on the same side of the metal center. At the
current level of theory, we find that the analogous structure with
C2V symmetry (H2 axis perpendicular to the ZrC2H2 plane) is a
saddle point with one imaginary frequency of 89i cm-1, lying
less than 1 kcal/mol above the minimum described here. We
will find additional evidence of soft hydrogen motion relative
to the ZrC2H2 frame further along the singlet reaction path. In
Figure 4. Details of RB3LYP/Stuttgart+6-311++G(d,p) optimized
geometries of potential minima and transition states for Zr + C2H4 the earlier study,13 neither the geometry nor the identity of the
along the singlet PES. Distances in angstroms, angles in degrees. For spin state of the H2ZrC2H2 complex were reported. Nonetheless,
each transition state, the transition vector corresponding to the imaginary based on its binding energy (Table 1), the species was surely
frequency is shown. the singlet dihydrido intermediate.
TABLE 3: Calculated Energies (kcal/mol) of Stationary Before discussing the elimination of H2 from the dihydrido
Points along the Singlet PESa intermediate (7) in some detail, we describe the ZrC2H2 (1A1)
+ H2 products themselves. The geometry and energy of the
RB3LYP/ UB3LYP/ RmPW1PW91/
Stuttgart+ Stuttgart+ Stuttgart+ ZrC2H2 (1A1) complex (8) are given in Figure 4 and Table 3.
species 6-311++G(d,p) 6-311++G(d,p) 〈Ŝ 〉 6-311++G(d,p)c
2 b
The structure of 8 is virtually identical to the strongly bound
ZrC2H4 (1A1) (5) -29.3 (-29.1) -31.2 (-31.0) 0.6 -35.2 triplet metallacyclopropene (4b). The binding energy of 48.5
1
TSins -7.9 (-6.8) -7.9 (-6.8) 0 -15.3 kcal/mol and the calculated reaction exothermicity of -8.3 kcal/
HZrC2H3 (6) -24.6 (-22.7) -24.6 (-23.3) 0 -31.9 mol are also close to the corresponding triplet energies. There
1TS
βH -19.9 (-17.9) -19.9 (-17.9) 0 -27.3
H2ZrC2H2 (7) -49.5 (-47.4) -49.5 (-47.4) 0 -54.3 is a significant admixture of the Zr (4d25s2, 1D) electronic state
1
TSexit -4.3 (-2.5) -12.2 (-10.2) 0.9 -7.1 (d1.59s1.71, Table 4), and the bonding mechanism follows the
ZrC2H2 (1A1) + -8.3 (-6.1) -13.9 (-11.6) 0.8 -10.1 description of the triplet metallacyclopropene outlined above
H2 (8) except that the two nonbonding electrons are spin-paired in an
a Energies relative to ground-state, triplet reactants, Zr (4d25s2, 3F) sd hybrid orbital. There is a considerable shift of out-of-plane
+ C2H4, corrected for differential zero-point energy effects. Parentheses π electron density (0.19 e-) into the empty dxz orbital, but no
denote energetics for Zr + C2D4. b Computed average value 〈Ŝ2〉 appreciable metal back-donation.
measures the spin contamination of the UB3LYP determinants. c Single-
point energies for structures optimized at the level of RB3LYP/ An important mechanistic question is whether a low-energy
Stuttgart+6-311++G(d,p). singlet reaction path leads from the dihydrido intermediate (7)
to elimination products. Hence, we now describe in detail our
dihydrido intermediate (7) is described in Figure 4 and Table investigation of coordinate scans along the singlet exit channel
3. Again, it is the simultaneous interaction of Zr in its singlet PES. The results are summarized in Figure 5, calculated at the
spin state with C1, C2, H1, and H4 that lowers the energy of level of RB3LYP/Stuttgart+6-311++G(d,p). The open triangles
1TS 3
βH relative to TSβH by 10 kcal/mol. The NBO analysis plot the energy (uncorrected for ZPE) as a function of decreasing
identifies β-CH agostic interactions with the metal center that the H1-Zr-H4 bond angle in H2ZrC2H2 (7). A geometry
account for the substantial lengthening of the C2-H1 bond (1.32 optimization is performed at each fixed value of H1-Zr-H4.
Å). Furthermore, the transition vector in Figure 4 involves The energy smoothly increases to +3.4 kcal/mol at H1-Zr-
Reaction Paths for Gas-Phase Zr + C2H4 J. Phys. Chem. A, Vol. 105, No. 20, 2001 4859

Figure 5. RB3LYP/Stuttgart+6-311++G(d,p) potential energy surface scans of the singlet exit channel connecting H2ZrC2H2 (7) to ZrC2H2 (1A1)
+ H2 products. At each value of angle H1-Zr-H4, all other coordinates are optimized. Energies uncorrected for zero-point effects. Open triangles
depict the energy as a function of increasing the H1-Zr-H4 bond angle from 7. Points A, B, and C in the inset mark points along this scan. Filled
squares illustrate the IRC following from 1TSexit. Point D marks the minimum found along this scan. Filled circles depict the energy as a function
of decreasing the H1-Zr-H4 bond angle from D. See text for details.

H4 ) 20° (A), abruptly decreases to +0.9 kcal/mol at H1- previous bond angle scan carried out in the opposite direction
Zr-H4 ) 19° (B), and then slowly rises again to a maximum (triangles). All the structures represented by the closed circles
at H1-Zr-H4 ) 16° (C). These positions are marked in the are planar.
inset of Figure 5. Accompanying the abrupt change in energy In effect, at this level of theory we have identified two
between A and B, there is but a small change in geometry. In diabatic surfaces that intersect in the vicinity of H1-Zr-H4
both product-like species, the structural parameters of the ) 22°. By diabatic we mean preserving both electron config-
ZrC2H2 unit are very similar to ZrC2H2 (1A1) (8, Figure 2). uration and spin. An NBO analysis of the Zr population at A
Furthermore, in A (B) the Zr-H1 and Zr-H4 distances are gives d2.61s0.51 and reveals that both the electron configuration
2.18 Å (2.33 Å) and 2.18 Å (2.25 Å), respectively, and the H1- of Zr (d1.93s0.45) and the nonplanar geometry of H2ZrC2H2 (7)
H4 distance is 0.757 Å (0.759 Å). The primary difference is have been retained along the forward bond angle scan (triangles).
the position of H2 relative to the ZrC2H2 frame. Figure 5 shows In contrast, the points along the IRC scan (squares) and the
views of A and B with the ZrC2H2 unit oriented almost reverse bond angle scan (circles) have electron configurations
perpendicular to the plane of the paper. A is nonplanar, while that more closely match Zr (d1.71s1.59) in ZrC2H2 (1A1) (8) and
B is planar. planar geometries. For example, the configurations at H2-
Using the structure at C (also planar) as input, a successful ZrC2H2 (D) and 1TSexit are Zr (d2.00s1.36) and Zr (d1.81s1.50),
transition state optimization was performed. The geometry and respectively. Qualitatively speaking, from the dihydrido inter-
energy of 1TSexit are shown in Figure 4 and Table 3; its energy mediate (7) toward products, electron density must flow out of
uncorrected for ZPE is also indicated in the inset of Figure 5. the ZrH bonds into a nonbonding orbital. Along the diabatic
Again, the planar transition state 1TSexit is very product-like. curve emanating from 7, the nonbonding electrons evolve to
The structural parameters of the ZrC2H2 unit are similar to 8; occupy an sd-hybrid orbital of mostly d-character, while the
the Zr-H1 and Zr-H4 distances have increased to 2.74 and electrons occupy an sd-hybrid of primarily s-character along
2.64 Å, respectively, and the H1-H4 distance is 0.752 Å. The the diabatic curve that correlates to products. Despite the similar
labeled transition vector depicts motion that clearly leads to geometries near the surface intersection, the energies differ
elimination products. The filled squares in Figure 5 illustrate because of the different orbital occupancies. However, along
the results of the IRC following from 1TSexit outward, confirm- the forward bond angle scan (triangles), the calculations do find
ing the connection to the ZrC2H2 (1A1) + H2 exit channel. In a lower energy solution to the SCF equations at B that directly
the direction back toward 7, the IRC scan identifies a shallow correlates to the ZrC2H2 (1A1) + H2 exit channel, involving a
minimum at H1-Zr-H4 ) 21° (D). The structure of that change of electron configuration to the d2s2-like state of the
weakly bound H2-ZrC2H2 complex is virtually identical to the products.
triplet H2-ZrC2H2 complex 3b, described above (Figure 2). The RB3LYP calculations are evidently unable to find the
In an attempt to establish the connection of the singlet H2- adiabatic surface that smoothly connects the dihydrido inter-
ZrC2H2 (D) complex to the strongly bound dihydrido intermedi- mediate (7) to elimination products. All attempts to find a
ate H2ZrC2H2 (7), we performed a potential energy surface scan transition state in the region of the curve-crossing failed, using
increasing the H1-Zr-H4 angle from the product-like mini- either wave function as the initial guess. Searches for the
mum. The filled circles in Figure 5 illustrate those results. We adiabatic reaction path involving other reaction coordinates were
find that the energy rapidly increases, crossing that of the also unsuccessful. For example, in scans involving the H4-
4860 J. Phys. Chem. A, Vol. 105, No. 20, 2001 Porembski and Weisshaar

H1-Zr-C2 dihedral angle of the H2-ZrC2H2 complex (D) and some bonding interaction with the π bond of the alkene or
scans using a radial coordinate (distance from Zr to the midpoint alkyne. As a result, the barrier to the addition of ground-state
of H2), we have located another set of geometries that fail to Zr to the CC bond is no more than 1 kcal/mol. Second, two
converge near the same surface intersection at essentially the distinct reaction paths on the triplet PES have been found, a
same total energy. As described above, the electron configu- stepwise path and a concerted path. Both ultimately lead to
ration at Zr must change substantially from H2ZrC2H2 (7) en exothermic ZrC2H2 (3A2) + H2 products. Although the quantita-
route to products. RB3LYP cannot provide sufficient multi- tive details depend on the density functional employed, all
configurational character in the vicinity of the diabatic surface methods predict that the lowest energy path involves the
intersection to accurately describe the nature of such a transition concerted elimination of H2 from the strongly bound insertion
state. intermediate (2b) over the multi-center transition state 3MCTS
Our results, however, clearly indicate that there does exist a with no subsequent exit channel barrier.
low-energy singlet path from 7 to products. Because the two Interestingly, for the H2 elimination mechanisms of Fe+ (3d6-
diabatic surfaces that intersect are of the same spin, the mixing 4s , 6D), Co+ (3d8, 3F) and Ni+ (3d9, 2D) with ethane and
1

and splitting at the intersection shown in Figure 5 should be propane, B3LYP calculations find that the lowest energy reaction
substantial. Furthermore, since the diabatic crossing point path again involves an MCTS.1,5-8 Other calculations have
already lies below 1TSexit, we expect no additional barrier along predicted the concerted elimination of H2 via a similar MCTS
the adiabatic surface. Thus, in Figure 3 and our mechanistic for the neutral transition metal reaction Pt (5d96s1, 3D) + CH4.30
discussion below, we consider 1TSexit to be the effective exit In all such examples, concerted rearrangement to a loosely
channel barrier to formation of the ZrC2H2 (1A1) + H2 products bound exit-channel complex over a multi-centered transition
from the dihydrido intermediate (7) on the singlet PES. state is energetically favored over stepwise rearrangement
Significantly, 1TSexit lies 4 kcal/mol above product when both because the metal center lacks sufficient bonding capacity to
are computed at the RB3LYP level. Since the geometry of form covalent bonds to both hydrogen atoms and to the
1TS remaining hydrocarbon fragment. Agostic interactions between
exit is already quite product-like, we expect that a more
flexible open-shell singlet treatment of both 1TSexit and singlet the metal center and the σCH bond or the σHH bond, as found
products would preserve this ∼4 kcal/mol barrier. here for 3MCTS, evidently play a significant role in the stability
of all such transition states.
IV. Discussion The new calculations also find a low-energy singlet path to
exothermic ZrC2H2 (1A1) + H2 products. All barriers associated
A. Overview of Experimental and Theoretical Results. with primary CH insertion and stepwise H2 elimination along
Chemical kinetics data at 300 K (0.9 kcal/mol mean collision the singlet PES lie well below ground-state triplet reactants,
energy) and 0.5-0.8 Torr buffer gas provide the bimolecular due to the ability of Zr in its singlet spin state to form strong
rate constant for Zr + C2H4, kH ) 55 × 10-12 cm3 s-1.12 The covalent bonds. This extra bonding capacity of singlet Zr
reaction efficiency, estimated as kH/kLJ, is only 7%; here kLJ is obviates the need for a MCTS. Therefore, in contrast to the
the Lennard-Jones rate constant estimated as before.4 No kinetic triplet PES, the Zr + ethylene reaction on the singlet PES
isotope effect is observed on perdeuteration; kH/kD ) 1.05 ( proceeds via the stepwise mechanism, involving a small exit
0.05.4 Over the limited flow tube pressure range of 0.5-1.1 barrier to the elimination of H2 from the strongly bound
Torr, PIMS at 157 nm detects only elimination products, ZrC2H2 dihydrido intermediate (7).
or ZrC2D2. Stabilized collision complexes are estimated to B. Reaction Mechanism. 1. Triplet Path ReVisited. The long-
contribute no more than 10% of the primary products, based standing view of the Zr + ethylene reaction based on the PCI-
on experimental sensitivity. 80 calculations of Siegbahn et al. (Table 1) involves rate-
Crossed-beam scattering angular distributions provide comple- determining primary CH insertion from the metallacyclopropane
mentary information on the Zr + C2H4 reaction at higher complex on the triplet spin surface.13 Other stationary points
collision energies between 6 and 23 kcal/mol.15 At 5.9 kcal/ along the stepwise H2 elimination path were found to lie well
mol, the center-of-mass angular distribution of nonreactive Zr below the energy of reactants, and any exit channel barrier from
atoms is distinctly forward peaked, indicating the absence of the strongly bound dihydrido intermediate was anticipated to
backscattered metal atoms from dissociation of long-lived be small. It is now clear that both the βH transfer transition
ZrC2H4 collision complexes. Evidence of such long-lived state and the H2ZrC2H2 complex were in fact stationary points
complexes is first observed at 9.1 kcal/mol, and becomes along the singlet spin surface.
substantial at both 14.0 and 23.1 kcal/mol. At all collision In a more recent study, UB3LYP/LANL2DZ calculations
energies, ZrC2H2 products are detected with forward-backward confirmed the mechanism of primary CH insertion along the
symmetric center-of-mass angular distributions and kinetic triplet PES outlined by the earlier theoretical work.4 Further-
energy distributions P(Et′) that peak distinctly above zero kinetic more, using RRKM theory31,32 based on the UB3LYP/
energy. Such product distributions are consistent with a mech- LANL2DZ geometries and frequencies, we modeled the decay
anism that proceeds via intermediates having lifetime greater of long-lived metallacyclopropane complexes and explored what
than a rotational period, and with a modest exit channel barrier range of energies of key stationary points yield behavior
above the ZrC2H2 + H2 product asymptote. consistent with experiment. Both the absence of an isotope effect
The electronic structure calculations of this study reveal many at 300 K4 and the absence of backscattered Zr at low collision
new mechanistic features of the Zr + ethylene reaction. First, energies15 indicate that CH insertion by the metallacyclopropane
the existence of the long-range complexes Zr-C2H4 (1a) and must be facile. Specifically, when 3TSins lies at -4 to -7 kcal/
Zr-C2H2 (4a) shows that rather than being chemically inert, s2 mol, essentially all complexes go on to insert into a CH bond
metal atoms can form chemical bonds with little or no barrier. of ethylene and eliminate H2 rather than dissociate back to
For the early transition metals Y and Zr, the fully occupied s reactants. Under room-temperature reaction conditions, the rate
orbital can hybridize with empty d orbitals to effectively reduce model predicts that ZrC2H4 complexes then live on the order
electron repulsion toward the hydrocarbon, while maintaining of 1 ns, too short for stabilization by a third-body collision in
Reaction Paths for Gas-Phase Zr + C2H4 J. Phys. Chem. A, Vol. 105, No. 20, 2001 4861

the flow tube at 0.5-1.1 Torr. This is consistent with the product propane well (1b). Next, some fraction of the triplet complexes
branching data from PIMS. At higher collision energy, angular undergo triplet-to-singlet intersystem crossing (ISC) at a rate
momentum barriers atop 3TSins deflect some ZrC2H4 complexes kISC, gaining access to the singlet metallacyclopropane well (5).
back to reactants, as experimentally observed.15 The kinetics The rate model for the triplet mechanism yields a rough bound
data set an upper bound on 3TSins, while the onset of backscat- on the time scale for which the spin transition must occur in
tering in turn sets a lower bound on the energy of 3TSins. order to compete with facile CH bond insertion on the triplet
Assuming all collisions follow the lowest adiabatic triplet surface. With energies of 3TSent (+0.78 kcal/mol) and 3TSins
surface, a rate-determining entrance channel barrier 3TSent (-6.1 kcal/mol) adjusted to fit the data as described above, the
(0.5-2 kcal/mol) to complex formation would then be necessary lifetime of triplet ZrC2H4 complexes with respect to CH bond
to limit the reaction efficiency, although no such barrier was insertion is about 1 ns at 300 K. Alternatively, in the extreme
found at the level of UB3LYP/LANL2DZ. Alternatively, limit of no insertion on the triplet surface, ISC must be
nonadiabatic effects along the triplet entrance channel could also sufficiently fast (∼10 ns) so that no more than about 5% of
contribute to the reaction inefficiency. Some of the potentials triplet ZrC2H4 complexes dissociate back to reactants at low
emanating from the 5-fold electronically degenerate triplet collision energy. In either event, ISC on a time scale of 1 ns or
ground-state Zr (4d25s2, 3F2) + C2H4 are likely repulsive. Only shorter would make the singlet mechanism competitive. Un-
a fraction of collisions may actually access the attractive fortunately, it is difficult to estimate a priori the absolute
adiabatic surface that leads to the formation of the metallacy- magnitude of kISC.33
clopropane complex and subsequent H2 elimination. If the singlet surface is reached, the present calculations
We must now reassess these conclusions in light of the new indicate that subsequent H2 elimination is also efficient, involv-
theoretical results that have clearly identified three different ing facile CH insertion by the metallacyclopropane (5) followed
reaction paths along two electronic spin surfaces. First, we by stepwise rearrangement to the very deep dihydrido interme-
consider the possibility that the Zr + ethylene reaction occurs diate well (7). From RRKM calculations on the singlet surface,
exclusiVely on the adiabatic triplet PES. We repeated the RRKM we find that even without downward adjustment the RB3LYP
calculations using the new UB3LYP/Stuttgart+6-311++G(d,p) energy of -7.9 kcal/mol for 1TSins is sufficiently low to account
rotational constants and vibrational frequencies for 3TSent, for the lack of a deuterium isotope effect and the absence of
ZrC2H4 (3A2) (1b), and 3TSins. Because we find that the backscattered Zr at low collision energies. Rearrangement of
stationary point geometries and frequencies are relatively the singlet insertion intermediate (6) to the dihydrido intermedi-
insensitive to the choice of basis set, neither the qualitative nor ate (7) is facile at all collision energies, since the barrier 1TSβH
quantitative results of the earlier statistical model change. The is so small. Reverse intersystem crossing from singlet H2ZrC2H2
calculated UB3LYP energy of 3TSent (+0.78 kcal/mol) is (7) back to the triplet surface is probably inefficient, since there
consistent with the experimentally determined reaction ef- is no triplet species with geometry or energy similar to 7.
ficiency. However, we must lower 3TSins by 8 kcal/mol from Therefore, we reason that those collisions that access the
the UB3LYP energy of +1.9 kcal/mol to -6.1 kcal/mol, much singlet surface are likely to eliminate H2 from H2ZrC2H2 (7) to
as before. Complexes then insert much faster than they produce the ZrC2H2 (1A1) + H2 products. In fact, the small exit
dissociate, resulting in no isotope effect and no backscattered barrier 1TSexit (4 kcal/mol for RB3LYP) is more consistent with
Zr at low collision energies; the centrifugal barrier atop 3TSins the peaking of the P(Et′) between 3 and 5 kcal/mol than is
scatters Zr back to reactants at the higher collision energies. 3MCTS or 3TS
exit on the triplet surface. Although the calculated
The triplet PES then branches into two reaction paths from the exothermicity of -8.3 kcal/mol for singlet products is smaller
insertion intermediate (2b), but the low-energy 3MCTS ensures than the estimate of -16 kcal/mol from the experimental P(Et′),
that the concerted mechanism dominates at all collision energies the RB3LYP calculations likely underestimate the binding
of interest here. energy of ZrC2H2 (1A1) (8), and hence the exothermicity of
Statistical modeling of the kinetic energy release distributions singlet products. The spin-restricted method also overestimates
is beyond the scope of this study, but qualitative comparison to the energy of 1TSexit. We emphasize that a more flexible open-
the product scattering data suggests that neither triplet mech- shell singlet treatment would likely preserve the key ∼4 kcal/
anism is consistent with the experimental distributions. The mol energy difference between 1TSexit and the ZrC2H2 (1A1) +
P(Et′) found to best fit the ZrC2H2 laboratory angular distribution H2 product asymptote because the geometry of 1TSexit is already
at a collision energy of 14.0 kcal/mol peaks at 3-5 kcal/mol so product-like. Finally, in the singlet mechanism the data still
and extends to about 30 kcal/mol, indicating the presence of a require that a small approach barrier or nonadiabatic effects
small exit barrier. The UB3LYP exothermicity of -14.9 kcal/ along the ground-state triplet entrance channel limit the reaction
mol for triplet products compares favorably with the estimate efficiency, as already described for the triplet mechanism.
of -16.0 kcal/mol from the maximum observed kinetic energy. 3. EVidence from Different DFT Methods. Table 1 shows that
However, along the lower energy concerted path, we find no for the triplet Zr + ethylene PES, the UB3LYP, UmPW1PW91,
exit barrier to elimination of H2 from H2-ZrC2H2 (3b). Even and UB1LYP functionals yield quite different energies at
in the unlikely event that the 1 kcal/mol of potential energy at corresponding stationary points. Relative to reactants, the
3MCTS is channeled entirely into translation, this triplet path
energies consistently lie in the order E(UB1LYP) > E(UB3LYP)
cannot account for the product P(Et′) that peaks at 4 ( 1 kcal/ > E(UmPW1PW91). At both transition states and potential
mol. Along the higher energy stepwise path, the exit barrier wells, UB3LYP energies lie 2-4 kcal/mol below UB1LYP
3TS
exit lies 18 kcal/mol above products, which would almost energies and UmPW1PW91 energies lie an additional 4-11
surely result in a much larger kinetic energy release than kcal/mol below UB3LYP energies. Similarly, Table 3 shows
observed experimentally. that the RmPW1PW91 energies are also lower by 5-7 kcal/
2. CompetitiVe Singlet Path. The present work raises the mol than the corresponding RB3LYP energies for most station-
strong possibility that a significant fraction of the reactive events ary points along the singlet PES.
occur on the singlet PES. In this singlet mechanism, we assume Ideally, we would use our statistical rate model to indicate
that all reactive collisions first reach the triplet metallacyclo- the acceptable range of key stationary point energies based on
4862 J. Phys. Chem. A, Vol. 105, No. 20, 2001 Porembski and Weisshaar

experimental data and thus help judge the accuracy of the

different DFT methods. However, in the Zr + ethylene reaction
the additional mechanistic possibilities introduced by discovery
of the low-lying singlet surface make it very difficult to choose
a preferred functional. Nevertheless, much experience has been
garnered concerning the performance of the different functionals
on other related systems. Based on the results of this study, it
is clear that if the reaction took place entirely on the triplet
PES, the rate-limiting 3TSins must lie 4-7 kcal/mol below
reactants, i.e., the UB3LYP energy must be lowered by 6-9
kcal/mol. This is only slightly more than the downward
adjustments of 5-7 kcal/mol systematically required of rate-
limiting multi-center transition states in analogous B3LYP Figure 6. Schematic of lowest energy triplet and singlet reaction paths
studies of Co+ and Ni+ reactions with alkanes.1-3,5 The based on mPW1PW91 energetics, including three parallel, competitive
UB1LYP energies are unrealistically high. decay rates from the triplet metallacyclopropane complex 1b.
In contrast, the UmPW1PW91 energies seem quite consistent
with experiment, predicting facile CH bond insertion on both data are summarized in Figure 6. The two surfaces shown are
the triplet and singlet surfaces. While placement of 1TSins at the lowest energy, concerted triplet path and the stepwise singlet
-15.3 kcal/mol precludes any backscattering of Zr once path described in detail above. The energetics along both paths
reactants reach the singlet surface, placement of 3TSins at -6.2 are those predicted by the mPW1PW91 functional (Tables 1
kcal/mol is consistent with the onset of backscattering of high and 3), since those results seem most realistic. Accordingly, in
angular momentum components for collision energy between the triplet entrance channel we draw a barrierless, attractive
6 and 9 kcal/mol. Since the ISC rate remains unknown, the surface and a representative repulsive surface to suggest that
UmPW1PW91 energies do not answer the question of triplet/ the 7% reaction efficiency measures the fraction of room-
singlet branching. However, additional support for UmPW1PW91 temperature Zr + ethylene collisions that access the attractive
energetics over UB3LYP energetics comes from our ongoing surface and reach the metallacyclopropane complex (1b). Along
study of the simpler Y + ethylene reaction.34 In that system, the triplet concerted path connecting the insertion intermediate
only a doublet electronic surface is important, allowing a (2b) to elimination products (4b), UmPW1PW91 actually lowers
3MCTS below the ZrC H (3A ) + H asymptote, so we draw
more critical comparison of theory with experiment. For Y + 2 2 2 2
C2H4, the UB3LYP functional again predicts CH bond insertion a monotonically rising energy path from 2b to 4b. From 1b,
to be rate-determining, but places 2TSins at +4.8 kcal/mol. the three rate constants in boxes represent the competition
RRKM calculations indicate that again we must lower the between CH insertion and subsequent H2 elimination on the
UB3LYP value by 6-9 kcal/mol to reproduce the experimental triplet surface (kins), dissociation back to triplet reactants (kdiss),
observations of a small positive isotope effect and a small and intersystem crossing to the singlet surface (kISC) and
fraction of stabilized YC2D4 collision complexes. In contrast, subsequent H2 elimination.
the UmPW1PW91 functional places 2TSins at -3.0 kcal/mol, The outcome of this competition likely depends on both total
in good agreement with the chemical kinetics data. internal energy E and complex angular momentum J. From
Returning to Zr + ethylene, another difficult issue is the cause RRKM theory, we expect kins(E,J) to increase exponentially with
of the reaction inefficiency. Either a small entrance channel E and to decrease rapidly with J (at fixed energy), since 3TSins
barrier exists, or a substantial fraction of collisions access is a tight transition state. We expect kdiss(E,J) to increase
repulsive excited-states at long range and fail to convert to the exponentially with E but to decrease more slowly with J, since
attractive potential leading to the triplet metallacyclopropane the triplet entrance channel transition state is loose (little or no
well (nonadiabatic effects). In the former picture, RRKM barrier). In contrast, from a wealth of experimental measure-
modeling requires that the 3TSent barrier height be about 1 kcal/ ments of nonradiative transition rates, we expect kISC to increase
mol, in excellent agreement with the values of 0.8 and 1.2 kcal/ only slowly with E and to be independent of J.33 The reason is
mol obtained by UB3LYP and UB1LYP, respectively. However, that the triplet complex (1b) and the singlet complex (5) have
both B3LYP and B1LYP are known to poorly model long-range very similar geometries and electronic energies. Intersystem
intermolecular forces, finding, for example, that the He2 and crossing rates in such near resonant cases can be very fast. For
Ne2 dimers are unbound near their van der Waals minima.11,20 example, modest-sized ketones exhibit S1 f T1 ISC rates of
The mPW1PW91 functional, which uses the same ratio of exact/ 109 s-1 or larger near the S1 origin.33 These rates increase
DFT exchange as B1LYP, but different exchange (mPW) and roughly linearly with excess energy, presumably due to the
correlation (PW91) functionals, was designed in part to improve modest electronic energy gap of ∼0.3 eV. By analogy, for the
the long-range behavior.19,20The fact that UmPW1PW91 finds triplet ZrC2H4 complex (1b) we expect kISC to be fast, especially
no triplet entrance channel barrier then suggests that the given the enhanced spin-orbit coupling in the 4d series.
inefficiency of the reaction is more likely due to nonadiabatic Ultimately, triplet-to-singlet ISC will compete with the rates
effects. This picture can also explain why Zr reacts nearly three kdiss and kins most effectively at low collision energy and for
times faster with propylene than with ethylene.12,4 Driven by high-J complexes at higher collision energy.
the larger density of rovibronic states, a larger fraction of Zr + The best experimental evidence for the singlet mechanism is
C3H6 collisions can make the requisite nonradiative transition the product kinetic energy distribution P(Et′) obtained at reactant
to the attractive adiabatic surface prior to scattering back to Et ) 14.0 kcal/mol. The distribution peaks between 3 and 5
reactants. kcal/mol, in accord with the calculated barrier on the singlet
4. Preferred Mechanism. Our best inferences for the Zr + (but not the triplet) surface. If singlet elimination is competitive
ethylene reaction mechanism gleaned from the new theoretical even at 14.0 kcal/mol, the singlet mechanism may well dominate
results, the statistical rate modeling, and all of the experimental in our 300 K kinetics experiments. While we cannot rule out
Reaction Paths for Gas-Phase Zr + C2H4 J. Phys. Chem. A, Vol. 105, No. 20, 2001 4863

some contribution from the triplet mechanism, we expect that We thank Prof. H. Floyd Davis for useful discussions.
P(Et′) on the triplet path would be statistical and peak very close Generous support of this research has come from the National
to zero energy, in contrast to the experimental distribution. Science Foundation (Grants CHE-9616724 and NSF-0071458)
Assuming the RmPW1PW91 energy of -15.3 kcal/mol for and the donors of the Petroleum Research Fund (Grant 33441-
1TS
ins is roughly correct, the rate model then indicates that the AC6). M.P. thanks the UW-Madison Department of Chemistry
onset of backscattering of Zr at the higher collision energies for a Martha Weeks graduate fellowship.
cannot be due to reactants that reach the singlet surface. Rather,
we suggest that at higher Et the deflection of high-J complexes Supporting Information Available: Rotational constants
from 3TSins on the triplet surface begins to overcome kISC, which and vibrational frequencies of all stationary points calculated
increases only slowly with energy. Although the product kinetic at the level UB3LYP/Stuttgart+6-311++G(d,p). Supporting
energy distributions for reactant Et ) 6-23 kcal/mol show no Information is available free of charge via the Internet at http:///
evidence of bimodality, there may well exist an energy regime pubs.acs.org.
in which both triplet and singlet insertion mechanisms contribute
to H2 elimination products. Because the calculated singlet exit References and Notes
barrier is small and the calculated energy difference between
(1) Yi, S. S.; Blomberg, M. R. A.; Siegbahn, P. E. M.; Weisshaar, J.
singlet and triplet products is at most 10 kcal/mol, electronic C. J. Phys. Chem. A 1998, 102, 395.
branching may not be apparent in the broad P(Et′) distributions (2) Blomberg, M. R. A.; Siegbahn, P. E. M.; Yi, S. S.; Noll, R. J.;
observed. Weisshaar, J. C. J. Phys. Chem. A 1999, 103, 7254.
(3) Reichert, E. L.; Yi, S. S.; Weisshaar, J. C. Int. J. Mass Spectrom.
V. Conclusions 2000, 196, 55.
(4) Porembski, M.; Weisshaar, J. C. J. Phys. Chem. A 2000, 104, 1524.
This comprehensive theoretical study provides a detailed view (5) Yi, S. S.; Reichert, E. L.; Holthausen, M. C.; Koch, W.; Weisshaar,
of the Zr + ethylene reaction, including explicit characterization J. C. Chem.-Eur. J. 2000, 6, 2232.
(6) Holthausen, M. C.; Koch, W. HelV. Chim. Acta 1996, 79, 1939.
of all reaction intermediates along both the triplet and singlet
(7) Holthausen, M. C.; Fiedler, A.; Schwarz, H.; Koch, W. J. Phys.
spin surfaces. Along the triplet PES, we have identified a new Chem. 1996, 100, 6236.
lowest-energy path involving the concerted elimination of H2, (8) Holthausen, M. C.; Koch, W. J. Am. Chem. Soc. 1996, 118, 9932.
but exclusiVe reaction along this path appears to be inconsistent (9) Becke, A. D. J. Chem. Phys. 1993, 98, 1372.
with the experimental data. Instead, we argue that primary CH (10) Koch, W.; Hertwig, R. H. Density functional theory applications
insertion and stepwise H2 elimination along the singlet PES must to transition metal problems. In Encyclopedia of Computational Chemistry;
Schleyer, P., Ed.; John Wiley: New York, 1998; Vol. 1, p 689.
play an important role in explaining the behavior of Zr + (11) Niu, S.; Hall, M. B. Chem. ReV. 2000, 100, 353.
ethylene, especially the product kinetic energy release. For the (12) Carroll, J. J.; Haug, K. L.; Weisshaar, J. C. J. Am. Chem. Soc.
first time, a survey of the different DFT methods strongly 1993, 115, 6962.
suggests that the mPW1PW91 functional better represents the (13) Carroll, J. J.; Haug, K. L.; Weisshaar, J. C.; Blomberg, M. R. A.;
energies of transition metal species compared to the widely used Siegbahn, P. E. M.; Svensson, M. J. Phys. Chem. 1995, 99, 13955.
B3LYP functional. (14) Wen, Y.; Porembski, M.; Ferrett, T. A.; Weisshaar, J. C. J. Phys.
Chem. A 1998, 102, 8362.
Despite this theoretical progress, additional experimental and (15) Willis, P. A.; Stauffer, H. U.; Hinrichs, R. Z.; Davis, H. F. J. Phys.
computational work is needed to clarify the relative contributions Chem. A 1999, 103, 3706.
of the triplet and singlet mechanisms as a function of reactant (16) Blomberg, M. R. A.; Siegbahn, P. E. M.; Svensson, M. J. Phys.
kinetic energy. Unfortunately, estimation of the key rate constant Chem. 1992, 96, 9794.
kISC is a difficult task.33 Detailed modeling of the product kinetic (17) Siegbahn, P. E. M.; Blomberg, M. R. A.; Svensson, M. J. Am.
Chem. Soc. 1993, 115, 1952.
energy distributions on both the triplet and singlet surfaces using (18) Siegbahn, P. E. M. Theor. Chim. Acta 1994, 87, 277.
phase space theory32,35 at 3MCTS and 1TSexit is now possible (19) Adamo, C.; Barone, V. Chem. Phys. Lett. 1997, 274, 242.
with the new electronic structure results as input. Product angular (20) Adamo, C.; Barone, V. J. Chem. Phys. 1998, 108, 664.
and kinetic energy distributions at lower collision energies would (21) Andrae, D.; Haussermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor.
be more sensitive to the magnitude of small exit barriers and to Chim. Acta 1990, 77, 123.
product singlet-triplet branching. It also remains important to (22) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A.; Stratmann,
establish experimentally the existence or absence of an approach R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K.
barrier, since this is a further distinguishing feature between N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
mPW1PW91 and B3LYP. A study of the Zr + ethylene reaction R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
using a variable-temperature flow reactor or a merged-beam Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz, J.
apparatus at very low collision energies (<1 kcal/mol) would V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.;
be quite revealing. Higher level, multi-configuration calculations Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng,
of the singlet exit channel PES can greatly improve the C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.;
Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon, M.;
description of the adiabatic singlet path and test whether the Replogle, E. S.; Pople, J. A. Gaussian 98, Revision A.6; Gaussian, Inc.:
barrier 1TSexit remains substantial. Pittsburgh, PA, 1998.
Similar theoretical and experimental studies of ground-state (23) Koch, W.; Holthausen, M. C. A chemist’s guide to density functional
theory; Wiley-VCH: New York, 2000.
Y and Nb with ethylene and propylene are now in progress.34,36
(24) Ziegler, T. Chem. ReV. 1991, 91, 651.
For Y, with only three valence electrons, we expect concerted (25) Gill, P. M. W. Density functional theory (DFT), Hartree-Fock
H2 elimination on the doublet surface to play an important (HF), and the self-consistent field. In Encyclopedia of Computational
mechanistic role. For Nb, much as for Zr, intersystem crossing Chemistry; Schleyer, P., Ed.; John Wiley: New York, 1998; Vol. 1, p 678.
from sextet reactants to the low-lying quartet surface is likely (26) Hay, P. J. J. Chem. Phys. 1977, 66, 4377.
to be important. The goal of this work must be to place the (27) Moore, C. E. NBS Circ. No. 467; U.S. Department of Commerce:
Washington, DC, 1949; Vol. I-III.
entire collection of experimental and theoretical results for the
(28) Glendening, E. D.; Badenhoop, J. K.; Reed, A. E.; Carpenter, J.
reactions of Y, Zr, and Nb with alkenes within a comprehensive E.; Weinhold, F. NBO 5.0, 5.0 ed.; Theoretical Chemistry Institute:
and consistent mechanistic picture. University of Wisconsin, Madison, 1998.
4864 J. Phys. Chem. A, Vol. 105, No. 20, 2001 Porembski and Weisshaar

(29) Schwenke, D. W.; Truhlar, D. G. J. Chem. Phys. 1985, 82, 2418. (33) Avouris, P.; Gelbart, W. M.; El-Sayed, M. A. Chem. ReV. 1977,
(30) Carroll, J. J.; Weisshaar, J. C.; Siegbahn, P. E. M.; Wittborn, C. 77, 793.
A. M.; Blomberg, M. R. A. J. Phys. Chem. 1995, 99, 14388. (34) Porembski, M.; Weisshaar, J. C. J. Phys. Chem. A, submitted.
(31) Gilbert, R. G.; Smith, S. C. Theory of Unimolecular and Recom- (35) Hanratty, M. A.; Beauchamp, J. L.; Illies, A. J.; van Koppen, P.
bination Reactions; Blackwell Scientific Publications: Oxford, U.K., 1990. A. M.; Bowers, M. T. J. Am. Chem. Soc. 1988, 110, 1.
(32) Baer, T.; Hase, W. L. Unimolecular reaction dynamics; Oxford (36) Porembski, M.; Weisshaar, J. C. J. Phys. Chem. A, work in
University Press: New York, 1996. progress.