You are on page 1of 8

583

ARTICLE
Chemical bonding in groups 10, 11, and 12 transition metal
homodimers — An electron density study
SeyedAbdolreza Sadjadi, Chérif F. Matta, and I.P. Hamilton
Can. J. Chem. Downloaded from www.nrcresearchpress.com by George Mason University on 06/24/13

Abstract: The properties of metal–metal bonding for transition metal homonuclear diatomics from groups 10, 11 and 12 are
studied within the framework of the quantum theory of atoms in molecules (QTAIM) at the coupled cluster CCSD and CCSD(T)
levels of theory. A novel approximate method developed by Keith and Frisch is used to augment electron densities calculated
with pseudopotentials with the missing relativistic core densities to obtain approximations to the total densities of the dimers.
The calculated delocalization indices for group 10 dimers are: Ni2 (1.6), Pd2 (0.44, an outlier in the group), and Pt2 (1.8); for group
11 dimers: Cu2 and Ag2 (1.01), and Au2 (1.13), all covalent bonds; for group 12: Zn2 (0.06), Cd2 (0.08), and Hg2 (0.09), all consistent
with weak van der Waals complexes. The picture of bonding obtained by examining the values of the electron density at the bond
critical points is consistent with the one obtained on the basis of these delocalization indices. A curious linear (instead of
exponential) dependence of the delocalization index on the electron density at the bond critical point is presented here as an
observation and will be investigated in more depth in later work. Several correlations between bond properties and bond
dissociation energies are also explored. It is found that, with the exception of the Ni2 dimer that exhibits considerable multi-
reference character, there are correlations between the calculated bond dissociation energies of the studied diatomics and
several bond critical point properties. These correlations are novel as they span a set of bonds between different pairs of
elements, while traditionally these correlations were reported for bonds between the same pair or elements but with different
substituents.
For personal use only.

Key words: topology of the electron density, chemical bonding in transition metals, quantum theory of atoms in molecules,
QTAIM, metal-metal bonding.

Résumé : Les propriétés des liaisons métal-métal pour les dimères homonucléaires des métaux de transition des groupes 10, 11
et 12 sont étudiés dans le cadre de la théorie quantique des atomes-dans-les-molécules (QTAIM) en faisant appel aux niveaux de
théorie CCSD et CCSD (T). Une nouvelle méthode développée par Keith et Frisch est utilisée pour augmenter des densités
électroniques calculées à l'entremise de pseudo potentiels avec les densités de cœur relativistes manquantes pour obtenir des
approximations aux densités totales des dimères. Les indices de délocalisation calculés pour les dimères du groupe 10 : Ni2 (1.6),
Pd2 (0.44, qui est en écart du reste du groupe) et Pt2 (1.8); pour les dimères du groupe 11 : Cu2 et Ag2 (1.01) et Au2 (1.13), toutes les
liaisons étant covalentes; et pour le groupe 12 : Zn2 (0.06), Cd2 (0.08) et Hg2 (0.09) indiquant de faibles liaisons du type van der
Waals. L'image qui émerge à propos de la nature de la liaison chimique obtenu en examinant les valeurs de la densité
électronique aux points critiques est complètement en accord avec celle obtenue sur la base des indices de délocalisation. Une
curieuse dépendance linéaire (au lieu d'exponentielle) de l'indice de délocalisation sur la densité électronique au point critique
est présenté ici comme une observation et sera enquêté en plus de profondeur dans un travail future. Plusieurs corrélations entre
les propriétés des liaisons chimiques et les énergies de leur dissociation sont aussi explorées. Il est constaté que, à l'exception du
dimer de Ni2, doté d'un caractère multiréférenciel considérable, il existe des corrélations entre les énergies de dissociation des
dimères étudiés et plusieurs des propriétés de leur point critiques de liaisons. Ces corrélations sont nouvelles car ils s'étendent
sur un ensemble de liaisons entre de paires d'éléments différentes, quand traditionnellement ces corrélations ont été trouvées
pour les liaisons entre une même paire d'élément, mais avec des substituents différents.

Mots-clés : la topologie de la densité électronique, la liaison chimique des métaux de transition, la théorie quantique des atomes
dans les molécules (QTAIM/TQADM), la liaison métal-métal.

Introduction vances over the past decade, and using software packages such as
Transition metal dimers have been studied for a number of DeMon, density functional theory (DFT) calculations are now fea-
years1 and continue to provide challenges for ab initio methods. sible for clusters of up to 100 transition metal atoms.3 However,
As noted by Yanagisawa, Tsuneda, and Hirao, nonhybrid density standard ab initio methods (which are only feasible for small
functionals typically give accurate bond distances but overesti- clusters) can be more accurate,4 and we employ coupled cluster
mate the bond dissociation energies, while hybrid density func- CCSD and CCSD(T) calculations5–8 for our studies of bonding in
tionals typically give accurate bond dissociation energies but transition metal homonuclear diatomics. These calculations, es-
overestimate bond distances.2 There have been significant ad- pecially when corrected to account for relativistic effects, are very

Received 21 December 2012. Accepted 9 February 2013.


S. Sadjadi. Department of Chemistry, Faculty of Science, University of Hong Kong, Hong Kong.
C.F. Matta. Department of Chemistry and Physics, Mount Saint Vincent University, Halifax, NS B3M 2J6, Canada; Department of Chemistry, Dalhousie University, Halifax,
NS B3H 4J3, Canada.
I.P. Hamilton. Department of Chemistry, Wilfrid Laurier University, 75 University Avenue West, Waterloo, ON N2L 3C5, Canada.
Corresponding authors: Chérif F. Matta (e-mail: cherif.matta@msvu.ca) and I.P. Hamilton (e-mail: ihamilton@wlu.ca).
This article is part of a Special Issue dedicated to Professor Dennis Salahub in recognition of his contributions to theoretical and computational chemistry.

Can. J. Chem. 91: 583–590 (2013) dx.doi.org/10.1139/cjc-2012-0549 Published at www.nrcresearchpress.com/cjc on 25 February 2013.
584 Can. J. Chem. Vol. 91, 2013

accurate for both bond distances and bond dissociation energies, Table 1. Zero-point energy (ZPE) corrections and
with few exceptions, such as Ni2 that exhibits a significant multi- basis set superposition error (BSSE) correction
reference character. with the Boys–Bernardi counterpoise (CP) method
Using the calculated molecular electron densities, empirically for all dimers in this study.
corrected for relativistic effects (vide infra), we employ Bader's ZPE CP
|⌬Ecorr |
quantum theory of atoms in molecules (QTAIM) to examine cor- Dimer Method (kcal/mol) (kcal/mol)
relations between bond properties and bond dissociation energies
for nine transition metal homonuclear diatomics belonging to Ni2 CCSD 0.52 1.02
groups 10, 11, and 12. The analysis of the electron densities accord- Pd2 CCSD 0.15 0.40
CCSD(T) 0.17 0.54
ing to QTAIM has been shown to shed invaluable light on the
Pt2 CCSD 0.37 29.26
nature and characterization of chemical bonding.9–19 Recently,
CCSD(T) 0.35 26.26
we have studied metal–metal bonding in the coinage metal homo- Cu2 CCSD 0.36 0.72
nuclear diatomics from the perspective of the analysis of their
Can. J. Chem. Downloaded from www.nrcresearchpress.com by George Mason University on 06/24/13

CCSD(T) 0.38 0.98


molecular electron densities, specifically, via analysis of elec- Ag2 CCSD 0.27 0.58
tronic properties at the bond critical points and the integrated CCSD(T) 0.27 0.69
atomic basin properties.20 In this work, an almost perfectly linear Au2 CCSD 0.26 0.70
correlation (r = 1.000) is reported between the electron density at CCSD(T) 0.26 0.84
the bond critical point (BCP), ␳b, and bond dissociation energy Zn2 CCSD 0.02 0.08
(BDE) of Cu2, Ag2, and Au2.20 The idea of correlating ␳b with bond Cd2 CCSD 0.02 0.10
order, and hence bond strength, of a bond between the same pair CCSD(T) 0.03 0.15
of elements in a series of related compounds is not new;21,22 how- Hg2 CCSD 0.02 0.16
ever, in the case of the coinage metal dimers, this correlation was CCSD(T) 0.03 0.21
CP
established for a case where the bond in question is between Note: |⌬Ecorr | = magnitude of counterpoise correction;
CP
different pairs of elements,20 hence its novelty. Other correlations |⌬Ecorr| = |(EA ⫺ ECP CP CP
A ) ⫹ (EB ⫺ EB )|; for A2 molecules: |⌬Ecorr| =

between descriptors of the metal–metal bonding in the coinage |2EA ⫺ 2ECP A |. CCSD(T) values for Zn 2 are not shown because
metal dimers and their respective BDE were also reported.20 In calculations failed to converge due to high multi-
reference character of the ground electronic state.
order of decreasing correlation, the BDE linear correlation coeffi-
cients are with (i) curvature normal to the bond path ␭1,2 (r =
−0.996), (ii) total energy density at the BCP (Hb) (r = −0.987), and (iii)
For personal use only.

old of 10−8 atomic units has been imposed on self-consistent field


logarithm of the delocalization index [log DI, or log ␦(A,B)] (r =
(SCF) density-matrix calculations. Thresholds of 1.5 × 10−5 and
0.925).20 It is noteworthy to mention that ␳b9 and ␦(A,B) have both
4.0 × 10−5 hartree/bohr for residual maximum forces and residual
been shown to correlate with the classical Lewis bond order.23,24
These statistics, however, include only 3 data points and, while root-mean square forces on the nuclei, respectively, were imposed
suggestive, are insufficient to claim general trends as we remark on all energy minimizations. Frequency calculations via har-
in Ref. 20 The purpose of this paper is to explore the generality of monic oscillator approximation were subsequently performed to
these preliminary findings by enlarging the molecular set to 9 obtain the zero point energy (ZPE).
metal dimers from groups 10, 11, and 12 (including the 3 coinage The total number of electrons and the number of electrons that
metal atoms dimers that were considered previously). The results have been replaced by the ECPnMDF pseudopotential for each
reflect the fundamental role played by the molecular electron element (total/ECPnMDF) are: Ni = 28/10, Pd = 46/28, Pt = 78/60,
density, a quantum mechanical observable, in the quantitative Cu = 29/10, Ag = 47/28, Au = 79/60, Zn = 30/10, Cd = 48/28, and
description of bonding, even in the case of the bond dissociation Hg = 80/60. Thus all electrons are replaced by the PP with the
energy, an energy difference that depends on the energies of both exception of the outer 18, 19, and 20 electrons (which are treated
reactants and products. explicitly) for groups 10, 11, and 12, respectively. For each metal
atom in these groups, 8 of the remaining electrons are consis-
Molecular set tently treated as core (outer core) and the rest as valence electrons
The diatomics considered in this study are the group 10, 11, and in our pseudorelativitic calculations. All calculated bond dissoci-
12 transition metal homonuclear dimers: Ni2, Pd2, Pt2, Cu2, Ag2, ation energies (BDE) were corrected for ZPE and for basis set su-
Au2, Zn2, Cd2, and Hg2. The ground electronic state of each dimer perposition error (BSSE) using Boys and Bernardi's counterpoise
was established by geometry optimization for both singlet and approach (CP).31,32 The results of these two corrections (ZPE, and
triplet states; the state exhibiting the lowest energy was taken as BSSE-CP) are collected in Table 1.
the ground electronic state. As discussed below, these species
exhibit a wide range of bonding, with bond dissociation energies Core-augmented pseudo-potential electron densities
that are very large for the group 10 dimers and very small for the Electron densities calculated using a pseudopotential, ␳PP, pres-
group 12 dimers. ent several problems for a topological analysis of the electron
density: (i) the lack of nuclear maxima and (ii) the appearance of
Computational details spurious topological features near the boundary of the core elec-
Electronic structure calculations trons replaced by the pseudopotential. An elegant approximation
Two sets of ab initio calculations have been carried out within introduced by Keith and Frisch (KF) provides a practical resolution
the Born–Oppenheimer and frozen-core approximations: (i) single of these problems.33
reference coupled cluster (CC) with single and double excitations In the KF approximation,33 Douglass–Kroll–Hess two-component
(CCSD) and (ii) CCSD with triple excitation correction CCSD(T). The (DKH2) scalar-relativistic34–37 all-electron DFT calculations were
combination of the nonrelativistic CCSD and CCSD(T) Hamiltoni- performed on every atom of the periodic table (in the free neutral
ans with the ECPnMDF pseudopotential (PP)25–28 and their corre- ground state) using a very large basis set. Each core subshell con-
sponding consistent aug-cc-pVTZ-PP(spdfg) basis set27–30 were tribution (␳ss) to the total electron density (␳total) was separately
constructed within the pseudorelativistic framework to account fitted to a linear combination of s-type (spherical) Gaussian func-
for both scalar relativistic and spin-orbit coupling effects between tions (␳ssfit). The total core electron density [␳total(core)] of a given
the inner core electrons and the valence electrons. A tight thresh- atom in a molecule is then approximated as:

Published by NRC Research Press


Sadjadi et al. 585

Table 2. Total energies of singlet and triplet states and calculated bond lengths and bond dissociation energies for group 10, 11, and 12
homonuclear diatomics and their corresponding experimental values. (Calculations at the CCSD/ECPnMDF-aug-cc-pVTZ-PP and CCSD(T)/
ECPnMDF-aug-cc-pVTZ-PP levels of theory, with the unrestricted formulation applied for open-shell species).
|⌬Esinglet-triplet| ⌬(R-Rexp) BDE BDEEXP ⌬(BDE-BDEexp)
Diatomic 2S+1 E(total) (a.u.) (kcal/mol) R (Å) Rexp (Å) (Å) (kcal/mol) (kcal/mol) (kcal/mol)
Ni2
RCCSD 1 −337.89882 2.1086
UCCSD 3 −337.94285 27.6 2.0443 2.15456±0.0004a −0.1103 28.87 47.0896 (0.0461)a −18.22
RCCSD(T) 1 Not foundⴱⴱ
UCCSD(T) 3 Not foundⴱⴱ
Pd2
RCCSD 1 −253.92234 18.8 2.8006 0.32 5.26 23.752c −18.49
UCCSD 3 −253.89244 2.3052 2.48b
Can. J. Chem. Downloaded from www.nrcresearchpress.com by George Mason University on 06/24/13

RCCSD(T) 1 −253.96383 10.3 2.7443 0.26 7.37 −20.38


UCCSD(T) 3 −253.94738 2.3516
Pt2
RCCSD 1 −237.70174 2.3234
UCCSD 3 −237.73301 19.6 2.3026 2.33297±0.00044d −0.0304 118.76 73.3031c 45.46
RCCSD(T) 1 −237.76185 2.3637
UCCSD(T) 3 −237.78504 14.6 2.3199 −0.0131 127.83 54.53
Cu2
RCCSD 1 −393.33686 39.6 2.2412 0.0219 38.89 47.9f −9.01
UCCSD 3 −393.27374 4.0692 2.2193e
RCCSD(T) 1 −393.39233 43.9 2.2213 0.0020 44.04 −3.86
UCCSD(T) 3 −393.32233 2.6931
Ag2
RCCSD 1 −292.92371 32.6 2.5562 0.0259 32.34 37.8g −5.46
UCCSD 3 −292.87170 3.9492 2.5303g
RCCSD(T) 1 −292.96396 36.4 2.5454 0.0151 36.35 −1.45
UCCSD(T) 3 −292.90603 3.3066
For personal use only.

Au2
RCCSD 1 −270.35686 62.9 2.5055 0.0340 43.79 53.2i −9.41
UCCSD 3 −270.25668 2.4723 2.4715h
RCCSD(T) 1 −270.39952 2.5002 0.0287 48.43 −4.77
UCCSD(T) 3 Not foundⴱⴱ
Zn2
RCCSD 1 −452.91603 65.3 4.4194 0.01 0.19 0.6919j, 0.7977k, (0.7448)ⴱ −0.55
UCCSD 3 −452.81192 2.5186 4.19j,k, 4.62l, (4.41)ⴱ
RCCSD(T) 1 Not found
UCCSD(T) 3 −452.84807 2.5096
Cd2
RCCSD 1 −334.55446 63.3 4.4018 0.34 0.30 0.9378m −0.64
UCCSD 3 −334.45351 2.8213 3.78 (0.03)m, 4.07k,
4.33l, (4.06)ⴱ
RCCSD(T) 1 −334.59159 65.2 4.0664 0.01 0.54 −0.40
UCCSD(T) 3 −334.48774 2.8186
Hg2
RCCSD 1 −305.92087 93.3 4.0667 0.43 0.44 1.086o, 1.164q, (1.125)ⴱ −0.67
UCCSD 3 −305.77221 2.7491 3.605n, 3.69o,
3.63p, (3.64)ⴱ
RCCSD(T) 1 −305.95984 94.7 3.8515 0.21 0.72 −0.41
UCCSD(T) 3 −305.80892 2.7519
Note: (a) Ref. 51. (b) Ref. 59. (c) Ref. 55. (d) Ref. 52. (e) Ref. 45. (f) Ref. 46. (g) Ref. 60. (h) Ref. 61. (i) Ref. 62. (j) Ref. 63. (k) Ref. 48. (l) Ref. 64. (m) Ref. 65. (n) Ref. 66. (o) Ref. 67.
(p) Ref. 49. (q) Ref. 68.
ⴱAverage value of experimental bond lengths.

ⴱⴱThe single reference calculations failed due to highly multi-reference character of the electronic state at the given multiplicity and/or very shallow potential well

on the PES which did not satisfy the tight criteria applied to optimize the geometry.

[1] ␳total(core) ≈ 兺
core
␳ssfit
The KF density so obtained thus incorporates scalar relativistic
effects in an approximate manner. The resulting approximate den-
sity is well-behaved, and all spurious critical points and pathological
subshells
features in the topology and topography of the density are removed,
The total (all-electron) molecular electron density is then ap- as demonstrated in the recent literature.20,33,38 The last equality in
proximated as: eq. [2] expresses the approximate electron densities analyzed in this

关 共 兲兴
paper as well as in our previous work.20
[2] ␳total ⫽ 兺
all atoms
␳total(core) ⫹ ␳PP ≈ 兺 兺
all atoms core
␳ssfit ⫹ ␳PP All electronic structure calculations in this paper were per-
formed using Gaussian 09-Rev.B0139 and all QTAIM calculations
subshells were performed using AIMAll.40
The Muller approximation41 to the two-electron density matrix
where, in the first two summations, “all atoms” means all atoms in terms of natural orbitals of the one-electron density matrix has
for which the core has been replaced by a PP. been used by AIMAll40 to calculate electronic delocalization prop-

Published by NRC Research Press


586 Can. J. Chem. Vol. 91, 2013

Table 3. Bond properties calculated at the CCSD/ECPnMDF-aug-cc-pVTZ-PP level of theory, with the unrestricted formulation applied for
open-shell species.
Group Species 2S+1 BDE Bond length ␦(A,B) ␳b ⵜ2␳b ␭1,2 ␭3 Vb Gb Hb
10 Ni2 3 28.9 2.0443 1.579 0.099 0.372 −0.110 0.592 −0.189 0.140 −0.049
10 Pd2 1 5.3 2.8006 0.439 0.035 0.097 −0.029 0.156 −0.038 0.031 −0.007
10 Pt2 3 118.8 2.3026 1.781 0.130 0.384 −0.128 0.640 −0.228 0.162 −0.066
11 Cu2 1 38.9 2.2412 1.014 0.063 0.146 −0.058 0.262 −0.079 0.058 −0.022
11 Ag2 1 32.3 2.5562 1.013 0.055 0.152 −0.045 0.243 −0.066 0.052 −0.014
11 Au2 1 43.8 2.5055 1.125 0.083 0.192 −0.074 0.341 −0.108 0.078 −0.030
12 Zn2 1 0.2 4.4194 0.057 0.002 0.003 −0.001 0.005 0.000 0.001 0.000
12 Cd2 1 0.3 4.4018 0.083 0.003 0.004 −0.001 0.006 −0.001 0.001 0.000
12 Hg2 1 0.4 4.0667 0.090 0.005 0.010 −0.002 0.015 −0.002 0.002 0.000
Can. J. Chem. Downloaded from www.nrcresearchpress.com by George Mason University on 06/24/13

Note: Dimensioned quantities are all in atomic units (a.u.) except bond lengths which are in Å. The subscript b signifies that the value is reported at the “bond
critical point”; ␦(A,B) is the delocalization index between atoms A and B; ␭n is the nth curvature of the electron density at the BCP where by convention |␭1| ≥ |␭2|; and
Vb, Gb, and Hb are the potential, gradient kinetic, and total energy densities at the BCP, respectively. Hb = Vb + Gb. The bond ellipticity, defined by ␧ = ␭1/␭2 − 1, is zero
for all these dimers that possess cylindrical symmetry (C∞-axis).

Table 4. Atomic properties calculated by numerical integration over atomic basins at the CCSD/ECPnMDF-aug-cc-pVTZ-PP level of theory, with
the unrestricted formulation applied for open-shell species.
Vol(⍀), Error
Diatomic N(⍀)a LI(⍀)a %loc(⍀)a 0.001a E(⍀)b ␮(⍀) 兺

E(⍀) (kcal/mol)
Ni2 18.000 17.219 95.66 250.170 −168.971102 0.333 −337.942203 0.41
Pd2 18.000 17.781 98.78 254.849 −126.961113 0.046 −253.922225 0.07
Pt2 18.000 17.109 95.05 292.433 −118.865524 0.286 −237.731051 1.23
Cu2 19.000 18.493 97.33 241.914 −196.668335 0.258 −393.336669 0.12
Ag2 19.000 18.493 97.33 282.114 −146.461704 0.287 −292.923710 0.00
Au2 19.000 18.437 97.04 285.031 −135.178061 0.233 −270.356859 0.00
Zn2 20.000 19.971 99.86 319.017 −226.458017 0.027 −452.916034 0.00
Cd2 20.000 19.958 99.79 368.526 −167.277232 0.044 −334.554465 −0.00
For personal use only.

Hg2 20.000 19.955 99.78 338.321 −152.960446 0.044 −305.920892 −0.01


aN(⍀) = total number of electrons in basin ⍀, excluding the ECP electrons; LI(⍀) = electron localization index; %loc(⍀) = percent localization = [LI(⍀)/N(⍀)] ×

100; Vol(⍀) 0.001 = atomic volume calculated up to the isodensity surface of ␳ = 0.001 a.u., i.e., the van der Waals surface.
bVirial-based total basin energy corrected by the molecular virial ratio. The last column (error) is the difference between the total energy obtained directly

from the electronic structure calculation and twice the atomic energy obtained by numerical integration, and hence, is a measure of the quality of the atomic
integration.

erties. Core electrons modeled by the ECP are not included in the The electronic ground state for Pd2 is predicted to be singlet,
calculation of the DIs. In general, DIs calculated with the Muller but the electronic ground state of this dimer has been determined
approximation41 (and which includes the effects of electron Cou- as 3 兺 u⫹ by photoelectron spectroscopy.54,55 This discrepancy
lombic correlation, approximately) are generally lower, often sig- between the theoretically and experimentally determined
nificantly, than their Hartree–Fock counterparts as Coulombic ground states for the group 10 dimers may be examined in
correlation disrupts electron pairing.42 future work.
In Table 2 the experimental bond lengths and bond dissociation
Results and discussions
energies are listed along with those calculated at the theoretical
Electronic ground states electronic ground states. Table 2 also lists the differences (⌬) be-
Considering the highly multiconfigurational character of some tween theory and experiment for the bond dissociation energies.
of the transition metal diatomics (in particular Ni2)43 and the The similar ⌬ values indicate that the CCSD and CCSD(T) results
significant effect of spin-orbit coupling on the ground electronic are consistent, and from this consistency it may be inferred that
structure of these dimers,44 it is necessary to establish the reliabil- information about the bonding from CCSD should be of almost
ity of the wavefunction in predicting the molecular properties the same quality as that from CCSD(T). This is important because
prior to a detailed topological analysis of the ensuing electron
at the time of writing the coupled cluster relaxed density matrix
densities.
can only be obtained from the CCSD wavefunctions.
The calculated CCSD and CCSD(T) total energies and their cor-
responding optimized internuclear distances (R) together with Correlation between bond dissociation energy (BDE) and
experimental internuclear distances (Rexp) are collected in Table 2. bond properties
For the group 11 and 12 dimers it may be seen that the singlet
A linear correlation between BDE and ␳b values was previously
states are lower in energy (more stable) than the triplet states,
reported for coinage metal dimers, where in all cases the bonding
with energy gaps greater than 30 kcal/mol (1 cal = 4.184 J). This
prediction of singlet electronic ground states for group 11 and 12 was characterized by p2␳b > 0, Hb < 0 and ␦(A,B) ≈ 1.0 at the MP2
dimers is consistent with the spectroscopically determined ground level.20 Here we examine this correlation at the CCSD level and
states.45–50 extend our study to the group 10 and 12 dimers.
In contrast, the electronic ground states for Ni2 and Pt2 are CCSD(T) calculations at the complete basis set limit and exper-
predicted to be triplet at both the CCSD and CCSD(T) levels, imental measurements (see BDE values in Table 2) provide con-
although spectroscopic experiments have suggested a mixture vincing evidence that group 12 dimers are all of the van der Waals
of singlet and triplet states for Ni251 and a singlet state for type.56 In contrast, recent ab initio calculations and subsequent
Pt2.52,53 population analyses of 3d and 4d transition metal dimers with

Published by NRC Research Press


Sadjadi et al. 587

Fig. 1. Correlation of BDE vs ␳b for homoatomic transition metal dimers. (a) Group 10. (b) Group 11. (c) Group 12. (d) All dimers pooled
together. (e) All dimers pooled together excluding Ni2. (Result of CCSD/ECPnMDF-aug-cc-pVTZ-PP calculations, with the unrestricted
formulation applied for open-shell species).
Can. J. Chem. Downloaded from www.nrcresearchpress.com by George Mason University on 06/24/13
For personal use only.

unfilled (n-1)d orbitals, such as Cr, W, and Mo, provide evidence of much lower DI linking the two Pd atoms is consistent with a much
multiple bonds.57 lower ␳b between these atoms (␳b = 0.035 a.u.), compared with a ␳b
Bond properties for the group 10 and 12 dimers and the corre- of 0.099 and 0.130 a.u. for Ni2 and Pt2, respectively. It is important
sponding atomic properties are presented in Tables 3 and 4, re- to underscore that all the studied features of metal bonding re-
spectively, all at the CCSD level. It may be seen that the bonding ported previously for the coinage metal dimers at the MP2 level
for the group 10 dimers is characterized by p2␳b > 0 and Hb < 0, are reproduced at the CCSD level qualitatively and quantitatively
indicating a location of the BCP in the valence shell charge deple- within reasonable accuracy (Tables 3 and 4).20
tion region together with a high contribution of potential energy The QTAIM bonding descriptors calculated for group 12 dimers
density (Vb) at this point (Table 3). These characteristics of bonding (Table 3) are consistent with and provide support to the picture of
are very similar to those previously reported for group 11 weakly bound dimers that emerged from experimental evidence
dimers.20 as described above.56 The bonding in this group is characterized
In contrast, a marked distinction between group 10 and group 11 by very small values of ␳b (0.002 to 0.005 a.u.), p2␳b > 0, ␦(A,B) < 0.1,
dimers is afforded by the values of the delocalization index (DI). and negative Hb of magnitudes less than 0.0005 a.u.
The values of the DI for group 10 dimers are substantially different Studies of the Laplacian of the electron density for group 11 and
than the corresponding values for group 11 dimers. The large 12 dimers have shown that the metal–metal bonding for transi-
values of ␦(Ni,Ni) = 1.579 and ␦(Pt,Pt) = 1.781 (for Ni2 and Pt2, re- tion metals is distinguished from that for main-group metals by a
spectively) indicate considerable electron sharing in excess of large valence shell charge depletion zone (VSCD). This zone be-
1½ electrons between the metal atoms; these values are equal to comes larger from top to bottom in these groups, and in particular
the number of electrons shared in the bonding, since there is no both Hg2 and Au2 show very large (and diffuse) VSCDs.58 Thus,
charge transfer in these dimers. We note in passing that a classical each of the three groups of transition metals chosen in this study
double bond nature of the Pt–Pt bond has been suggested on the is representative of one of the three fundamental types of bond-
basis of a comparison of the rotational constant values of Pt2 with ing, i.e., DI > 1.5 for group 10 (with the exception of Pd2), DI < 1.0
those of Au2.52 On the other hand, ␦(Pd,Pd) = 0.439 for Pd2, a for group 11, and weak van der Waals type bonding with DI < 0.1
bonding clearly different than its two congeners Ni and Pt. The for group 12.

Published by NRC Research Press


588 Can. J. Chem. Vol. 91, 2013

Fig. 2. Correlation of BDE with common QTAIM bond properties for all studied homoatomic transition metal dimers of the three groups
pooled together (the data points for Ni2 are shaded in gray). (a) DI [␦(A,B)]. (b) p2␳b. (c) ␭1,2. (d) ␭3. (e) Vb. (f) Hb. (Result of CCSD/ECPnMDF-aug-cc-
pVTZ-PP calculations, with the unrestricted formulation applied for open-shell species.)
Can. J. Chem. Downloaded from www.nrcresearchpress.com by George Mason University on 06/24/13
For personal use only.

Plots of BDE versus ␳b are shown in Fig. 1 for group 10, 11, and 12 ␭3, Vb, and Hb = Vb + Gb) is now examined (Table 3 and Fig. 2).
dimers. It may be seen that the BDE–␳b plots for the group 11 Figure 2 shows that the bond dissociation energies of all dimers
(Fig. 1b) and 12 (Fig. 1c) dimers are close to linear (r = 0.94 and 0.99, except Ni2 (the only clear outlier) exhibit close to linear functional
respectively). The BDE–␳b plot for the group 10 dimers (Fig. 1a) dependence with respect to each of the studied bonding descrip-
shows the largest deviation from linearity (r = 0.86) as well as the tors.
largest standard deviation. This large deviation is most probably The correlation between BDE and DI is of particular interest,
due to the multireference character of Ni2 and appears to indicate
since the delocalization index is correlated to the classical bond
that this particular dimer is not well represented by the level of
order, particularly in molecules with no charge transfer across the
calculation in this work.
The observation of a linear correlation within each group (10, 11, studied bond. Data in Table 3 can be fitted to the relationship
and 12) suggests that all of these dimers may lie on one general proposed by Matta and Hernández:24
BDE–␳b line. Pooling all calculated BDEs versus the corresponding
␳b values for all dimers in Fig. 1d clearly has a general linear trend, [3] ln[␦(A,B)] ⫽ a(␳b ⫺ b) ⫽ a␳b ⫺ c
with the Ni2 constituting the most noticeable outlier. The linear
correlation coefficient of the regression line in Fig. 1d is 0.876, where a, b, and c are fitting parameters, a having dimensions of
slightly higher than that for group 10. In Fig. 1d, the calculated
reciprocal electron density and b having those of the density. This
BDE values versus ␳b values for all dimers are replotted after the
omission of Ni2, which results in a significant improvement of the relation between the DI and ␳b, originally proposed for carbon–
linear correlation (r = 0.953). Substitution of the experimental carbon bonds, has been found to also apply for the metal–metal
BDE value for Ni2 yields a value of ␳b = 0.067 a.u., a prediction to be bonding in the three coinage metal dimers.20 Fig. 3 shows direct
tested with future multireference calculations. relation between ␦(A,B) and ␳b, which in the case of the studied
To provide a more complete picture of metal–metal bonding for dimers exhibits no obvious curvature as expected from the expo-
group 10, 11, and 12 transition metal homonuclear diatomics, the nential dependence, ␦(A,B) = exp[a(␳b − b)], a curious feature pre-
correlation of BDE with other bonding indices (␦(A,B), p2␳b, ␭1,2, sented here as an observation.

Published by NRC Research Press


Sadjadi et al. 589

Fig. 4. BDE plotted against the natural logarithm of the DI (ln[␦(A,B)]) for homoatomic transition metal dimers. (a) Group 10. (b) Group 11.
(c) Group 12. (d) All dimers pooled together but excluding Ni2. (Result of CCSD/ECPnMDF-aug-cc-pVTZ-PP calculations, with the unrestricted
formulation applied for open-shell species.)
Can. J. Chem. Downloaded from www.nrcresearchpress.com by George Mason University on 06/24/13
For personal use only.

Fig. 3. Plot of the DI (␦(A,B)) as a function of ␳b for all homoatomic ison with several other common QTAIM bonding indices), despite
transition metal dimers including Ni2. (Result of CCSD/ECPnMDF- the fact that in this work the bonds included in the statistics are
aug-cc-pVTZ-PP calculations, with the unrestricted formulation different in nature and do not share a common element.
applied for open-shell species.)
Conclusions
We have employed coupled cluster CCSD and CCSD(T) calcula-
tions with the frozen-core approximation in combination with
recently developed pseudopotential-based correlation consistent
basis sets, followed by relativistic core augmentation, to obtain
electron densities for nine group 10, 11, and 12 transition metal
homonuclear diatomics. Using the calculated electron densities,
Bader's quantum theory of atoms in molecules (QTAIM) has been
used to calculate the bond and atomic basin properties of these
dimers. Features of metal–metal bonding observed previously for
coinage metal dimers (specifically, Hb < 0 and p2␳b > 0) were also
observed for group 10 and 12 dimers. The calculated ␳b and ␦(A,B)
values show that the nature of the metal–metal bonding in these
three groups of transition metal dimers is markedly different. In
traditional chemistry language, dimers are characterized by par-
tial and multiple bonds for group 10, by single bonds for group 11,
The BDE–␦(A,B) correlation plots are shown in Fig. 4a–4c for and by van der Waals interactions for group 12, a picture consis-
group 10, 11, and 12 dimers. The linear correlation coefficients r for tent with our calculations.
the group 10 and 11 dimers are significantly less than 1.00 (r = 0.76 In spite of the fundamental differences in metal–metal bond-
and 0.82, respectively) in view of the marked curvature of the ing, we have shown that for all transition metal dimers (with the
functional dependence of BDE on the logarithm of the delocaliza- exception of Ni2 that exhibits considerable multireference char-
tion index. The value of r for the coinage metal dimers, 0.826, is acter) there is a linear correlation between the strength of the
lower than that previously reported at the MP2 level (r = 0.925) metal–metal bond and the value of the electron density at the
reflecting the almost equal DI values for Cu2 and Ag2 at the CCSD
bond critical point (and also with several other QTAIM bonding
level of theory.20 The linear correlation coefficient for the group
descriptors). This correlation has previously been established for
12 dimers is closer to unity (r = 0.905). Figure 4d depicts the calcu-
lated BDE versus DI values for all dimers except Ni2. The linear bonding between the same pair of elements in a series of related
correlation coefficient for this plot is significantly less than unity compounds. We have established this correlation for bonding be-
(r = 0.774) and much lower than that for the corresponding BDE–␳b tween different pairs of elements. Whether this correlation is con-
plot (r = 0.95) in Fig. 1e. fined to certain transition metal dimers or is general for all
These results are consistent with previous work that has shown diatomics (homo and hetero, metal and non-metal) remains an
that BDE is correlated with ␳b in almost linear fashion (in compar- open question.

Published by NRC Research Press


590 Can. J. Chem. Vol. 91, 2013

Acknowledgements (33) Keith, T. A.; Frisch, M. J. J. Phys. Chem. A 2011, 115, 12879. doi:10.1021/
jp2040086.
The authors dedicate this paper to Professor Denis Salahub. (34) Jansen, G.; Hess, B. A. Phys. Rev. A 1989, 39, 6016. doi:10.1103/PhysRevA.39.
This research was conducted using the HKU Information Technol- 6016.
ogy Services Research Computing Facilities and Mount Saint Vin- (35) Hess, B. A. Phys. Rev. A 1986, 33, 3742. doi:10.1103/PhysRevA.33.3742.
cent University computer cluster and has been supported (36) Hess, B. A. Phys. Rev. A 1985, 32, 756. doi:10.1103/PhysRevA.32.756.
(37) Douglas, M.; Kroll, N. M. Ann. Phys. (N. Y.) 1974, 82, 89. doi:10.1016/0003-
financially by Hong Kong UGC Special Equipment Grant (SEG 4916(74)90333-9.
HKU09), Natural Sciences and Engineering Research Council of (38) Bendeif, E.-E.; Matta, C. F.; Stradiotto, M.; Fertey, P.; Lecomte, C. Inorg. Chem.
Canada (NSERC), Canada Foundation for Innovation (CFI), and 2012, 51, 3754. doi:10.1021/ic2026347.
Mount Saint Vincent University. (39) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.;
Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.;
References Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda,
(1) Salahub, D. R. Ab Initio Methods in Quantum Chemistry Part 2 - (Advances in R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.;
Chemical Physics, Volume 69); Lawley, K. P., Ed.; Wiley-Interscience, John Wiley Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.;
Can. J. Chem. Downloaded from www.nrcresearchpress.com by George Mason University on 06/24/13

and Sons, New York, 1987; pp. 447–520. Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.;
(2) Yanagisawa, S.; Tsuneda, T.; Hirao, K. J. Chem. Phys. 2000, 112, 545. doi:10. Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.;
1063/1.480546. Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.;
(3) Koster, A. M.; Geudtner, G.; Calaminici, P.; Casida, M. E.; Dominguez, V. D.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev,
Flores-Moreno, R.; Gamboa, G. U.; Goursot, A.; Heine, T.; Ipatov, A.; O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.;
Janetzko, F; del Campo, J. M.; Reveles, J. U.; Vela, A.; Zuniga-Gutierrez, B.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.;
and Salahub, D. R. DeMon2k, Version 3; The DeMon Developers, Cinvestav, Dapprich, S.; Daniels, A. D.; Farkas, Ö.; Foresman, J. B.; Ortiz, J. V.;
Mexico City, 2011. Cioslowski, J.; Fox, D. J. Gaussian 09, Revision B.01; Gaussian Inc., Wallingford
(4) Diaconu, C. V.; Cho, A. E.; Doll, J. D.; Freeman, D. L. J. Chem. Phys. 2004, 121, CT, 2010.
10026. doi:10.1063/1.1798992. (40) Keith, T. A. Available from http://Aim.Tkgristmill.Com/, 2011.
(5) Shavitt, I.; Bartlett, R. J. Many-Body Methods in Chemistry and Physics: MBPT and (41) Müller, A. M. K. Phys. Lett. A 1984, 105, 446. doi:10.1016/0375-9601(84)91034-X.
Coupled-Cluster Theory; Cambridge University Press, Cambridge, 2009. (42) Wang, Y.-G.; Matta, C. F.; Werstiuk, N. H. J. Comput. Chem. 2003, 24, 1720.
(6) Levine, I. N. Quantum Chemistry; 6th ed.; Pearson Prentice Hall, Upper Saddle doi:10.1002/jcc.10435.
River, New Jersey, 2009. (43) Jiang, W.; DeYonker, N. J.; Wilson, A. K. J. Chem. Theor. Comput. 2011, 8, 460.
(7) Szabo, A.; Ostlund, N. S. Modern Quantum Chemistry: Introduction to Advanced doi:10.1021/ct2006852.
Electronic Structure Theory; Dover Publications, Inc., New York, 1989. (44) Cheskidov, A. V.; Buchachenko, A. A.; Bezrukov, D. S. J. Chem. Phys. 2012, 136,
(8) Cramer, C. J. Essentials of Computational Chemistry: Theories and Models; John 214304. doi:10.1063/1.4721624.
Wiley and Sons, Ltd., New York, 2002. (45) Ram, R. S.; Jarman, C. N.; Bernath, P. F. J. Mol. Spectr. 1992, 156, 468. doi:10.
(9) Bader, R. F. W. Atoms in Molecules: A Quantum Theory; Oxford University Press, 1016/0022-2852(92)90247-L.
Oxford, UK, 1990. (46) Rohlfing, E. A.; Valentini, J. J. J. Chem. Phys. 1986, 84, 6560. doi:10.1063/1.
For personal use only.

(10) Popelier, P. L. A. Atoms in Molecules: An Introduction; Prentice Hall, London, 450708.


2000. (47) Krämer, H. G.; Beutel, V.; Weyers, K.; Demtröder, W. Chem. Phys. Lett. 1992,
(11) Matta, C. F.; Boyd, R. J. The Quantum Theory of Atoms in Molecules: From Solid 193, 331. doi:10.1016/0009-2614(92)85639-R.
State to DNA and Drug Design; Wiley-VCH, Weinheim, 2007. (48) Czajkowski, M. A.; Koperski, J. Spectrochim. Acta A 1999, 55, 2221. doi:10.1016/
(12) Bader, R. F. W. J. Phys. Chem. A 2009, 113, 10391. doi:10.1021/jp906341r. S1386-1425(99)00020-7.
(13) Bader, R. F. W. J. Phys. Chem. A 1998, 102, 7314. doi:10.1021/jp981794v. (49) van Zee, R. D.; Blankespoor, S. C.; Zwier, T. S. J. Chem. Phys. 1988, 88, 4650.
(14) Tsirelson, V. G.; Ozerov, R. P. Electron Density and Bonding in Crystals: Principles, doi:10.1063/1.453777.
Theory and X-ray Diffraction Experiments in Solid State Physics and Chemistry; (50) Koperski, J.; Atkinson, J. B.; Krause, L. Chem. Phys. Lett. 1994, 219, 161. doi:10.
Institute of Physics Publishing, New York, 1996. 1016/0009-2614(94)87039-X.
(15) Koritsanszky, T. S.; Coppens, P. Chem. Rev. 2001, 101, 1583. doi:10.1021/ (51) Pinegar, J. C.; Langenberg, J. D.; Arrington, C. A.; Spain, E. M.; Morse, M. D.
cr990112c. J. Chem. Phys. 1995, 102, 666. doi:10.1063/1.469562.
(16) Coppens, P. X-ray Charge Densities and Chemical Bonding; Oxford University (52) Airola, M. B.; Morse, M. D. J. Chem. Phys. 2002, 116, 1313. doi:10.1063/1.
Press, Inc., New York, 1997. 1428753.
(17) Macchi, P.; Proserpio, D. M.; Sironi, A. J. Am. Chem. Soc. 1998, 120, 13429. (53) Fabbi, J. C.; Langenberg, J. D.; Costello, Q. D.; Morse, M. D.; Karlsson, L.
doi:10.1021/ja982903m. J. Chem. Phys. 2001, 115, 7543. doi:10.1063/1.1407273.
(18) Bader, R. F. W.; Matta, C. F.; Cortés-Guzmán, F. Organometallics 2004, 23, (54) Ho, J.; Ervin, K. M.; Polak, M. L.; Gilles, M. K.; Lineberger, W. C. J. Chem. Phys.
6253. doi:10.1021/om049450g. 1991, 95, 4845. doi:10.1063/1.461702.
(19) Bader, R. F. W.; Matta, C. F. Inorg. Chem. 2001, 40, 5603. doi:10.1021/ic010165o. (55) Ho, J.; Polak, M. L.; Ervin, K. M.; Lineberger, W. C. J. Chem. Phys. 1993, 99,
(20) Sadjadi, S.; Matta, C. F.; Lemke, K. H.; Hamilton, I. P. J. Phys. Chem. A 2011, 115, 8542. doi:10.1063/1.465577.
13024. doi:10.1021/jp204993r. (56) Pahl, E.; Figgen, D.; Borschevsky, A.; Peterson, K.; Schwerdtfeger, P. Theor.
(21) Grimme, S. J. Am. Chem. Soc. 1996, 118, 1529. doi:10.1021/ja9532751. Chem. Acc. 2011, 129, 651. doi:10.1007/s00214-011-0912-1.
(22) Exner, K.; Schleyer, P. v. R. J. Phys. Chem. A 2001, 105, 3407. doi:10.1021/ (57) Roos, B. O.; Borin, A. C.; Gagliardi, L. Angew. Chem. Int. Ed. 2007, 46, 1469.
jp004193o. doi:10.1002/anie.200603600.
(23) Fradera, X.; Austen, M. A.; Bader, R. F. W. J. Phys. Chem. A 1999, 103, 304. (58) Sadjadi, S. Ph.D. thesis, Hong Kong University, Hong Kong, 2012.
doi:10.1021/jp983362q. (59) Morse, M. D. Chem. Rev. 1986, 86, 1049. doi:10.1021/cr00076a005.
(24) Matta, C. F.; Hernández-Trujillo, J. J. Phys. Chem. A 2003, 107, 7496. doi:10. (60) Ran, O.; Schmude, R. W., Jr.; Gingerich, K. A.; Wilhite, D. W.;
1021/jp034952d; (a) (Correction: J. Phys. Chem A 2005, 109, 10798. doi:10.1021/ Kingcade, J. E., Jr. J. Phys. Chem. 1993, 97, 8535. doi:10.1021/j100134a025.
jp055864r). (61) Huber, K. P.; Herzberg, G. Molecular spectra and molecular structure IV. Constants
(25) Dolg, M. Theor. Chem. Acc. 2005, 114, 297. doi:10.1007/s00214-005-0679-3. of Diatomic molecules; Van Nostrand, Princeton, 1979.
(26) Figgen, D.; Rauhut, G.; Dolg, M.; Stoll, H. Chem. Phys. 2005, 311, 227. doi:10. (62) James, A. M.; Kowalczyk, P.; Simard, B.; Pinegar, J. C.; Morse, M. D. J. Mol.
1016/j.chemphys.2004.10.005. Spect. 1994, 168, 248. doi:10.1006/jmsp.1994.1275.
(27) Peterson, K. A. F. D.; Dolg, M.; Stoll, H. J. Chem. Phys. 2007, 126, 124101. (63) Strojecki, M.; Ruszczak, M.; Krośnicki, M.; Łukomski, M.; Koperski, J. Chem.
doi:10.1063/1.2647019. Phys. 2006, 327, 229. doi:10.1016/j.chemphys.2006.04.008.
(28) Figgen, D.; Peterson, K. A.; Dolg, M.; Stoll, H. J. Chem. Phys. 2009, 130, 164108. (64) Ceccherini, S.; Moraldi, M. Chem. Phys. Lett. 2001, 337, 386. doi:10.1016/S0009-
doi:10.1063/1.3119665. 2614(01)00203-2.
(29) Peterson, K. A.; Puzzarini, C. Theor. Chem. Acc. 2005, 114, 283. doi:10.1007/ (65) Strojecki, M.; Krośnicki, M.; Zgoda, P.; Koperski, J. Chem. Phys. Lett. 2010, 489,
s00214-005-0681-9. 20. doi:10.1016/j.cplett.2010.02.039.
(30) Peterson, K. A. Personal communication, 2012. (66) Koperski, J.; Qu, X.; Meng, H.; Kenefick, R.; Fry, E. S. Chem. Phys. 2008, 348,
(31) Boys, S. F.; Bernardi, F. Mol. Phys. 1970, 19, 553. doi:10.1080/ 103. doi:10.1016/j.chemphys.2008.02.035.
00268977000101561. (67) Koperski, J.; Atkinson, J. B.; Krause, L. Can. J. Phys. 1994, 72, 1070. doi:10.1139/
(32) van Duijneveldt, F. B.; van Duijneveldt-van de Rijdt, J. G. C. M.; p94-139.
van Lenthe, J. H. Chem. Rev. 1994, 94, 1873. doi:10.1021/cr00031a007. (68) Wüstenbecker, G. Ph.D. thesis, Phillips University Marburg, Marburg, 2000.

Published by NRC Research Press

You might also like