Iron versus Ruthenium: Clarifying the Electronic Differences between Prototypical Mixed-Valence Organometallic Butadiyndiyl Bridged Molecular Wires

Article abstract of DOI:10.1021/acs.organomet.8b00099

The electronic structures of the prototypical bimetallic buta-1,3-diyn-1,4-diyl-bridged radical cation complexes [{M(dppe)Cp′}₂(μ-C≡CC≡C)]⁺ (M= Fe, Cp′ = Cp? (1a), Cp (1b); M = Ru, Cp′ = Cp? (2a), Cp (2b)) have been (re)investigated using a combination of UV-vis-NIR and IR spectroelectrochemistry, and quantum chemical calculations based on both dispersion-corrected global (BLYP35-D3) and local (Lh-SsirPW92-D3) hybrid functionals. Following analysis of new and existing data, including the IR-active ν(C≡C) bands, the iron compounds [1]⁺ are reclassified as valence-trapped (Robin and Day Class II) mixed-valence complexes, in contrast to the ruthenium complexes [2]⁺, which are delocalized (Robin and Day Class III) systems. All members of the series exist as a thermally populated distribution of conformers in solution, and the overlapping spectroscopic profiles make the accurate extraction of the parameters necessary for the analysis of [1]⁺ and [2]⁺ within the framework of the Marcus-Hush model extremely challenging. Analysis of the spin-density distributions from a range of conformational minima provides an alternative representation of the degree of charge localization, and a comparison between members of the series is presented.

Full text of DOI:10.1021/acs.organomet.8b00099

Article
Cite This: Organometallics XXXX, XXX, XXX−XXX
Iron versus Ruthenium: Clarifying the Electronic Differences between
Prototypical Mixed-Valence Organometallic Butadiyndiyl Bridged
Molecular Wires
Simon Guckel,† Josef B. G. Gluyas,‡ Sarah El-Tarhuni,∥ Alexandre N. Sobolev,‡,§ Mark W. Whiteley,∥
̈
Jean-Franco
̧
is Halet,⊥ Claude Lapinte,⊥ Martin Kaupp,*,† and Paul J. Low*,‡
†Institut fur Chemie, Technische Universitat Berlin, Sekr. C7, Strasse des 17. Juni 135, 10623 Berlin, Germany
‡School of Molecular Sciences and §Centre for Microscopy Characterisation and Analysis, University of Western Australia, 35 Stirling
Highway, Crawley, Perth, Western Australia 6009, Australia,
∥School of Chemistry, University of Manchester, Oxford Road, Manchester, M13 9PL, United Kingdom
⊥Universite
́
Rennes, CNRS, ISCR - UMR 6226, F-35000 Rennes, France
̈
̈
S
* Supporting Information
ABSTRACT: The electronic structures of the prototypical bimetallic buta-1,3-diyn-1,4-
diyl-bridged radical cation complexes [{M(dppe)Cp′}2(μ-CCCC)]+ (M= Fe, Cp′ =
Cp* (1a), Cp (1b); M = Ru, Cp′ = Cp* (2a), Cp (2b)) have been (re)investigated using
a combination of UV−vis−NIR and IR spectroelectrochemistry, and quantum chemical
calculations based on both dispersion-corrected global (BLYP35-D3) and local (Lh-
SsirPW92-D3) hybrid functionals. Following analysis of new and existing data, including
the IR-active ν(CC) bands, the iron compounds [1]+ are reclassified as valence-trapped
(Robin and Day Class II) mixed-valence complexes, in contrast to the ruthenium
complexes [2]+, which are delocalized (Robin and Day Class III) systems. All members of
the series exist as a thermally populated distribution of conformers in solution, and the
overlapping spectroscopic profiles make the accurate extraction of the parameters
necessary for the analysis of [1]+ and [2]+ within the framework of the Marcus−Hush
model extremely challenging. Analysis of the spin-density distributions from a range of
conformational minima provides an alternative representation of the degree of charge
localization, and a comparison between members of the series is presented.
Chart 1. Bimetallic Half-Sandwich Buta-1,3-diyn-1,4-diyl-
Derived Complexes under Discussion
INTRODUCTION
■
The syntheses, redox properties, molecular and electronic
structures of metal complexes bearing linear, “all-carbon”
ligands, [{LmM}(μ-C4){MLm}]n+ have been widely investigated
since the first examples were synthesized in the mid 1970s.1
Among these investigations, the elucidation of the structure of
the carbon-ligand and nature of the “mixed-valence” derivatives
(n = 1) as a function of the metal, M, and supporting ligands,
Lm, have been topics of considerable interest.2−22 Four of the
early examples of [{LmM}(μ-CCCC){MLm}]+ complexes,
specifically [{Cp*(dppe)Fe}(μ-CCCC){Fe(dppe)Cp*}]+
([1a]+) (Chart 1),2,3 [{Cp*(dppe)Ru}(μ-CCCC){Ru-
(dppe)Cp*}]+ ([2a]+) (Chart 1),10 [{Cp(PPh3)2Ru}(μ-C
CCC){Ru(PPh3)2Cp}]+,9 and [{Cp*(NO)(PPh3)Re}(μ-
density at the metal from the ν(NO) band in the case of
[{Cp*(NO)(PPh3)Re}(μ-CCCC){Re(PPh3)(NO)-
Cp*}]+), analysis of the NIR band envelopes found in the
∼5000−15 000 cm−1 region within the framework of Marcus−
Hush theory, supported by quantum-chemical calculations
using HF or standard global hybrid functionals (B3LYP), were
6,23
CCCC){Re(PPh3)(NO)Cp*}]+
have become estab-
lished as benchmark examples of highly delocalized or Class III
mixed-valence complexes. Each complex is comprised of a buta-
1,3-diyn-1,4-diyl fragment terminated by half-sandwich, d6
metal fragments and exhibits characteristic ν(CC) and NIR
spectra. While the IR spectra provide information concerning
the formal bond order and a valence-bond description of the C4
ligand (augmented by information concerning the electron
Received: February 19, 2018
© XXXX American Chemical Society
A
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
used to provide information concerning electronic structure in
these earlier works.5−7,9,11
structure that occur as a function of redox state. The
electrochemical responses of many complexes of the general
formula {(η-C5R5)(PP)M}(μ-CCCC){M(PP)(η-C5R5)}
(M = Fe, Ru) have been reported and discussed on previous
occasions.1,5 In general, the complexes undergo three (M =
Fe)4 or four (M = Ru)8−10,14,16 oxidation processes within the
electrochemical window of common solvents. New compound
1b displays three redox processes in 0.1 M NBu4PF6/CH2Cl2
electrolyte at a Pt electrode (−0.94, −0.28, +0.53 V vs Fc/Fc+ =
0.00 V, Fc*/Fc*+ = −0.55 V Figure S2), which are as reversible
as the internal decamethylferrocene/decamethylferrocinium
reference couple, paralleling the behavior of the Cp* analogue
1a (−1.14, −0.42, +0.49 V) under identical conditions.
Comparison of these data reveals trends which support and
extend concepts developed in earlier studies of various
members of the {(η-C5R5)(dppe)M}(μ-CCCC){M-
(dppe)(η-C5R5)} series (M = Fe, Ru; R = Me, H).11 First,
the formal substitution of Cp (R = H) for Cp* (R = Me)
causes a more significant shift to less positive (more negative)
potentials of the first (ΔE1 = −0.20 V) and second (ΔE2 =
−0.14 V) redox processes than the third (ΔE3 = −0.04 V).
Second, the third redox potentials are almost independent of
the nature of supporting cyclopentadienyl ligand (E3 = +0.95 V
(1a), +0.99 V (1b)). This is entirely consistent with the notion
that the latter redox events are more carbon-chain in
character,4,9 which is supported here by the observation of a
low-frequency ν(CCCC) band by in situ spectroelectrochem-
istry for [1b]3+ (1684 cm−1).
Recent re-investigations of bimetallic ruthenium complex 2a
(Chart 1), supported by analysis of closely related complexes
including [{Cp(dppe)Ru}(μ-CCCC){Ru(dppe)Cp}]
(2b) (Chart 1), have shown that a more complete analysis of
these systems requires explicit consideration of thermal
populations of conformers that differ through rotation of one
half-sandwich fragment relative to the other around the Ru−
CC bond and hence offer different orbital overlaps over the
RuC4Ru chain.14,16 Conformational isomers have also been
detected in 31P NMR spectra of iron complex 1a, with the
rotational barrier determined to be some 28.2 kJ (6.74 kcal)
mol−1.24 The stability of the different rotamers, the thermal
population of which will be important to the observed
spectroscopic properties of solutions of the complex, are likely
influenced by a combination of both steric and electronic
interactions between the Fe(dppe)Cp* fragments. On the basis
of these observations, we were drawn to re-examine the
electronic structure and the electronic and vibrational
spectroscopic data of [{Cp*(dppe)Fe}(μ-CCCC){Fe-
(dppe)Cp*}]+ ([1a]+) and the closely related species [{Cp-
(dppe)Fe}(μ-CCCC){Fe(dppe)Cp}]+ ([1b]+) (Chart 1)
using specially constructed global and new local hybrid density
functionals. The new analysis allows us to assign key
spectroscopic features and to (re)classify [1a,b]+ in terms of
a ground state population distribution that features many
minima with significantly polarized (“localized”) electronic
structures. This significant ground state polarization in the
population of the iron systems affords a clear contrast between
the iron and ruthenium homologues. These studies further
highlight the difficulty in applying conventional methods of
analysis based on the Marcus−Hush theory to conformationally
fluxional “mixed-valence” systems of low symmetry.
Guided by the electrochemical results, the synthesis of [1b]n+
(n = 1 or 2) by chemical oxidation with ferrocenium
hexafluorophosphate was investigated. Oxidation of a cooled
(0 °C) dichloromethane solution of 1b with 2 equiv of
ferrocenium hexafluorophosphate resulted in a rapid color
change from orange to deep green, and the dication [1b][PF6]2
was subsequently isolated as a deep green solid. The
monocation [1b][PF6] could also be obtained by ferrocenium
oxidation of 1b, but the preferred protocol involves the
comproportionation reaction of equimolar quantities of [1b]-
[PF6]2 and 1b in CH2Cl2 which led to the isolation of
[1b][PF6] as a black-green solid. The oxidized complexes
[1b]n+ (n = 1 or 2) were characterized by IR spectroscopy,
mass spectrometry, and microanalytical data. IR data in the
ν(CC) stretching region recorded on isolated samples of
[1b]n+ fully concur with the data obtained from spectroelec-
trochemical investigations (see below). Microanalytical data
indicate the inclusion of some dichloromethane of crystal-
lization in both [1b][PF6] and [1b][PF6]2; a similar
observation was made for the Cp* analogue [1a][PF6].3
Samples of both [1b](PF6) and [1b](PF6)2 decomposed
slowly in air to give [Fe(CO)(dppe)Cp](PF6), characterized by
a strong ν(CO) band (1978 cm−1 (CH2Cl2); 1970 cm−1
(ATR)) and molecular ion at m/z (MALDI-MS) = 547 amu.
RESULTS AND DISCUSSION
■
The compound [{Cp(dppe)Fe}(μ-CCCC){Fe(dppe)-
Cp}] (1b) was prepared in the same two-step method as
developed earlier for the preparation of the well-known
complex [{Cp*(dppe)Fe}(μ-CCCC){Fe(dppe)Cp*}]
(1a), through a sequence of reactions involving initial oxidation
of Fe(CCH)(dppe)Cp to give the radical cation [Fe(C
CH)(dppe)Cp]+ and homodimerization to give the bis-
(vinylidene) [{Fe(dppe)Cp}2(μ-CC(H)C(H)C)]2+,25 fol-
lowed by deprotonation.3 Multinuclear NMR spectra for
butadiyndiyl complex 1b were obtained at room temperature
in the presence of a trace of CoCp2 to prevent accumulation of
Fe(III) species by aerial oxidation; this procedure is an
alternative to the low-temperature protocol employed for the
equivalent characterization of 1a.3 In the 13C{1H} NMR
spectrum, the quaternary carbons of the C4 bridge were
observed at δ 110.2 (Cβ) and δ 88.9 (t, JC−P = 44 Hz); the high-
field shift of Cα by comparison with related monometallic
alkynyl complexes [Fe(CCR)(dppe)Cp] is also a feature of
the analogous Fe(dppe)Cp* system. The room-temperature
31P{1H} NMR spectrum of 1b exhibits a singlet resonance at δ
105.0. Crystallographic characterization of 1b (Figure S1)
confirmed the butadiyndiyl form of the C4 bridge (Fe−C(α)
1.884(3) Å, C(α)−C(β) 1.220(4) Å, C(β)−C(γ) 1.373(4) Å,
C(γ)−Fe 1.889(3) Å), typical of complexes of this
type.5,11,26−30
An authentic sample of the carbonyl cation [Fe(CO)(dppe)-
31,32
Cp]+
was independently synthesized and isolated as the
¯
BPh4 salt, and shown to exhibit identical IR spectroscopic
(ν(CO) = 1979 cm−1 (CH2Cl2), 1970 cm−1 (ATR)) and mass
spectrometric properties to those of the decomposition product
from [1b](PF6)n. It has also been shown that in the Cp* series,
compounds with general formula [Fe(CCR)(dppe)Cp*]+
react readily with O2 to form the analogous carbonyl cation
[Fe(CO)(dppe)Cp*]+.33 Therefore, to avoid complications
arising from aerial decomposition of the chemically prepared
and isolated redox products, NIR and IR spectra of the redox
As noted above, much of the interest in butadiyndiyl-bridged
complexes stems from the changes in molecular and electronic
B
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
series [1a]n+ and [1b]n+ (n = 0, 1, 2) were collected in a
spectroelectrochemical cell of Hartl design (Table S1).34
Strictly anaerobic conditions were maintained during the
experiments by performing sample preparation, filling and
sealing the cell in a dinitrogen-filled dry glovebox using
rigorously dried and degassed solvents. Failure to observe these
precautions led to observation of the characteristic [Fe(CO)-
(dppe)Cp′]+ ν(CO) bands, indicating sample decomposition.
Before commencing a description of the experimental
spectra, it is helpful to first note the vibrational modes of a
model buta-1,3-diyne. The two highest frequency modes will
arise from the “symmetric” and “asymmetric” stretches of the
four-carbon fragment; the atomic displacements that corre-
spond to these modes are shown schematically in Figure 1. For
apparent differences in the IR spectra of these formally mixed-
valence monocations [1a,b]+ and [2a,b]+ can be summarized in
terms of the intensity of the higher-frequency, broad
“symmetric” ν(CCCC) band (Figure 2). The significant
IR activity of the “symmetric” ν(CCCC) band in [1a,b]+
points strongly to a noncentrosymmetric structure (i.e., valence
localization on the IR time-scale) of these iron complexes, while
the vanishingly small intensity of this vibration in [2a,b]+ points
to a more centrosymmetric (valence-delocalized) structure for
the ruthenium homologues.
This conclusion of a greater degree of localized (or Class II
mixed-valence) character in the iron compounds is seemingly at
odds with the greater separation of the first and second redox
waves associated with the oxidation of [1a] (ΔE1/2 = 0.72 V)
when compared with those of [2a] (ΔE1/2 = 0.65 V) and hence
larger comproportionation constant and thermodynamic
stability of [1a]+ (KC = 1.6 × 1012) compared to that of
[2a]+ (KC = 9.7 × 1010). However, the results of a relatively
recent analysis of the magnetic behavior of [1a][PF6]2 and
[2a][PF6]2 and thermodynamic terms associated with the
comproportionation equilibria give further insight,35 and we
begin to combine the various earlier observations and those
made here into a consistent picture. As discussed in various
degrees of detail in summaries published elsewhere,36−38 the
thermodynamic stability of mixed-valence complexes, expressed
as the free energy of comproportionation, ΔGC, can be
attributed to contributions from not only the “resonance”
effects that relate to the concepts of “metal−metal coupling”
but also entropic, electrostatic, inductive, solvation, ion-pairing,
and magnetic terms. From the temperature dependence of the
chemical shifts versus 1/T, a singlet−triplet energy gap of ΔGST
= −850 cm−1 was determined for [2a]2+, which is very much
greater than that found for [1a]2+ (ΔGST = −18 cm−1). The
comproportionation equilibrium, described as
Figure 1. Schematic representation of the ν(CCCC) modes in
buta-1,3-diyne.
a centrosymmetric molecule featuring the linear C4 fragment,
Y−CCCC−Y, the symmetric stretch will be IR inactive,
the asymmetric IR active. In contrast, in an asymmetrically
substituted derivative, X−CCCC−Y, both bands will be
IR active. Thus, for HCCCCH, a single (unscaled) IR-
active but low-intensity ν(CCCC) harmonic vibrational
frequency is calculated (BLYP35-D3/def2-SVP) at 2158 cm−1,
while HCCCCPh gave IR active ν(CCCC) modes at
2211 and 2376 cm−1 at the same level of theory.
In the IR spectra of 1a and 1b, the “asymmetric” ν(C
CCC) stretch is observed near 1955 (1a)24 or 1950 (1b)
cm−1 with discernible shoulders to the low (1a) or high (1b)
frequency side of the band (Figure 2). As has been discussed
elsewhere for related ruthenium butadiyndiyl complexes,14,16
these shoulders are characteristic of a distribution of con-
formers in solution; such rotamers have also been detected by
low-temperature 31P NMR spectroscopy in the case of 1a.24
The formally forbidden symmetric ν(CCCC) band has
been detected as a very weak feature near 2103 cm−1 (Figure
S3) in the IR spectrum of 1a, likely due to symmetry breaking
from rotation of the half-sandwich fragments around the long
molecular axis.24
[{LmM}(μ‐C≡ CC≡ C){MLm}]
+ [{LmM}(μ‐C≡ CC≡ C){MLm}]2+
⇆ 2[{LmM}(μ‐C≡ CC≡ C){MLm}]+
shifts to the left with greater stabilization of the (singlet)
dication [2a]2+, despite the greater resonance contribution
calculated previously from the Marcus−Hush two-state
model.35 In addition, while Mossbauer spectroscopic data
̈
from [1a]+ are consistent with a delocalized structure,3
Mossbauer operates on slower timescales (ca. 10−9 s) than
̈
molecular vibrations observed by IR measurements (ca. 10−13
s), suggesting a range for the possible electron transfer rate.
Further oxidation to the dications leads to another decrease
in the frequencies of the symmetric and asymmetric stretches of
the C4 bridge, with the symmetric band being almost IR-silent
and only detectable in the [M(dppe)(η-C5H5)] derived
complexes as an extremely weak feature near 1880 cm−1
([1b]2+) and 1860 cm−1 ([2b]2+) (Figure 2). The asymmetric
stretch was observed more clearly as a weak to medium
intensity band in the iron complexes at 1813 ([1a]2+) and 1807
([1b]2+) cm−1, close to those observed in the analogous
ruthenium complexes [{(η-C5R5)(dppe)Ru}(μ-CCC
C){Ru(dppe)(η-C5R5)}](PF6)2 (R = Me, 1770 cm−1
([2a]2+);10,16 R = H, 1772 cm−1 ([2b]2+),16 and taken as
evidence for a significant cumulene contribution to the bridging
C4 ligand structure.
The “mixed-valence” monocations [1a]+ and [1b]+ are
characterized by two main vibrational features corresponding to
the symmetric and asymmetric ν(CCCC) stretches at
1976 and 1880 cm−1 ([1a]+) and 1979 and 1877 cm−1 ([1b]+),
overlying a low-intensity electronic absorption band, and
providing the first clear evidence of valence-localization in at
least some of the conformations of these compounds on the IR
time scale. Each main vibrational feature also exhibits shoulders
at lower frequency, which point to additional conformational
isomers in solution. Close scrutiny of the IR spectra of the
isostructural ruthenium complexes [2a]+ and [2b]+ reveals a
very weak band feature near 2000 cm−1 for the symmetric
ν(CCCC) stretch in addition to the more prominent
bands for the asymmetric ν(CCCC) stretch at 1860
([2a]+) and 1859 ([2b]+) cm−1 (Figures 2 and S4). The
C
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Figure 2. Spectroelectrochemically generated IR ν(CC) spectra of [1a]n+, [1b]n+, [2a]n+, and [2b]n+ (n = 0, 1, 2) in 0.1 M NBu4PF6/CH2Cl2.
The electronic structures of [2a,b]+ have been explored
recently using computations with the global BLYP35-D3 hybrid
functional39−41 for a range of thermally accessible molecular
conformations that differ through the relative positions of the
half-sandwich fragments across the C4 bridge (Figure 3).14,16 In
each conformer, the compounds offer highly delocalized
electronic structures and may be described as Class III
mixed-valence complexes, albeit with a small polarization of
electron density in the “perpendicular” conformer.
It is likely that the small symmetry breaking arising from the
relative orientations of the Ru(dppe)Cp′ fragments in the perp
forms is responsible for the weak symmetric ν(CCCC)
band observed in the spectra of [2a,b]+ near 2000 cm−1 (Figure
2).16 Consistent with the lack of rotational dynamics in solid
samples, the ATR spectrum of [2a]PF6 shows a more
prominent signal corresponding to the symmetric stretch at
1992 cm−1 with the asymmetric stretch at 1854 cm−1 (Figure
S4). A significant increase in the relative intensity of the
Figure 3. Schematic representations of three low-lying minima of
symmetric ν(CCCC) feature relative to that seen in
[2b]+ for illustrative purposes.
solution is also apparent in solid-state spectra of [1a]PF6
(Figure S4). However, it seems unlikely that conformational
factors alone can satisfactorily explain the different intensities of
To better explain the IR spectra of [1a,b]+, computations at
the symmetric IR ν(CCCC) band of the iron and
BLYP35-D3/def2-SVP level were carried out. The 35% exact-
exchange admixture of BLYP35-D3 has been found to provide a
ruthenium species.
D
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
a
Table 1. Computed Harmonic Vibrational Frequencies for Different Conformers of [1a]+ and [1b]+ (BLYP35-D3/def2-SVP)
[1a]+
IR frequencies
[1b]+
IR frequencies
Ω [°]
ν1 (cm−1
)
int.
ν2 (cm−1
)
int.
Ω [°]
ν1 (cm−1
)
int.
ν2 (cm−1
)
int.
36
51
1953
1955
1971
1955
1952
1980
4376
4425
3709
4635
4999
1875
1876
1883
1876
1876
1877
168
333
179
258
321
55
1980
1987
1985
1969
1976
3451
3129
3126
3808
1888
1891
1890
1882
1880
41
44
22
51
94
86
110
144
exp
135
147
a
ν1, ν2 denote the symmetric and asymmetric alkynyl stretch, respectively. The computed frequencies are scaled by a factor of 0.895.43
Table 2. Comparison of Relative Energies (kJ mol−1) and Total Spin Expectation Values of Different Conformers of Mixed-
a
Valence Cations at Global (BLYP35-D3) and Local (Lh-SsirPW92-D3) Hybrid DFT Levels
BLYP35-D3
rel. E [kJ/mol]
Lh-SsirPW92-D3
rel. E [kJ/mol]
BLYP35-D3
rel. E [kJ/mol]
Lh-SsirPW92-D3
rel. E [kJ/mol]
Ω [°]
⟨S2⟩
⟨S2⟩
Ω [°]
⟨S2⟩
⟨S2⟩
[1a]+
[2a]+
0.77
0.77
0.77
0.77
0.77
0.77
[2b]+
0.77
0.77
0.77
0.77
0.77
0.77
36
51
2.1
0.0
2.1
0.0
0.76
0.76
0.76
0.76
0.76
0.76
55
94
0.0
1.7
3.4
8.3
0.83
0.83
0.83
0.83
0.0
8.8
4.8
6.0
0.79
0.79
0.79
0.79
69
4.6
11.8
25.5
22.9
20.4
110
144
105
132
153
20.2
16.4
13.8
[1b]+
0.82
0.82
0.82
0.82
0.82
36
51
0.0
3.0
0.0
10.3
20.6
17.6
18.8
0.79
0.79
0.79
0.79
0.79
33
51
0.0
2.6
0.0
8.3
0.76
0.76
0.76
0.76
0.76
0.76
86
8.5
98
21.0
24.7
22.6
23.3
23.0
31.2
34.0
32.6
135
147
17.2
17.0
127
150
174
a
At BLYP35-D3 structures. With COSMO solvent model (CH2Cl2) for both structure optimizations and single-point runs.
good compromise for modeling both localized and delocalized
organic, inorganic and organometallic mixed-valence systems.42
A conformational scan of the computationally less intensive
compound [1b]+ was undertaken, with sample points for initial
geometry optimizations of [1a]+ being drawn from those
identified for structurally similar [2a]+.16 Geometry optimiza-
tions were carried out giving four energetically similar minima
which are denoted by the torsion angle P−Fe···Fe−P (Ω°) for
[1a]+ (Ω = 55, 94, 110, and 144°) and five for [1b]+ (Ω = 36,
51, 86, 135, and 147°) (Table 1, Figure S5). The conforma-
tional distribution and dihedrals of these minima are
comparable to those of respective ruthenium homologues
[2a]+ (Figure S6) and [2b]+ (Figure S7). The lowest-energy
structures of [1a]+ and [2a]+ exhibit Ω close to 50° (55 and
51°, respectively), while [1b]+ and [2b]+ with the bulkier Cp*
ligands have values close to 30° (36 and 33°, respectively). As
discussed elsewhere for the ruthenium species,16 dispersion
interactions between the two ligand spheres are responsible for
these preferences and are strongest for the lowest-energy
minima. The observation of the critical role that dispersion
plays in these structures is intriguing, especially in light of the
IR spectra of the related compound [{Cp(dippe)Fe}(μ-C
CCC){Fe(dippe)Cp}]+ ([1c]+) featuring the smaller, alkyl
diphospine, di-isopropylphosphinoethane (dippe). In the solid
state as a Nujol mull, the IR spectrum of [1c]PF6 exhibits a
pattern of two ν(CC) bands (1977, 1879 cm−1) similar to
that observed for [1a]+ and [1b]+, while in CH2Cl2 solution,
only a single, asymmetric stretch ν(CC) band is observed at
1878 cm−1.4 It is interesting to speculate on the role that the
steric and intramolecular electronic (CH−π) interactions play
in influencing the conformational distribution in the entire
family of compounds [{LmM}(μ-CCCC){MLm}]n+,
although further explicit discussion of this point must await a
more comprehensive study of analogues of [1c]n+.
Earlier computations on truncated-ligand models for
7,11
[1a]+
gave delocalized structures and spin-density distribu-
tions. However, the low exact-exchange admixture (20%) of
B3LYP is known to clearly over-delocalize Class II systems.42
To provide a lower localization limit to the present
computations using full ligand models with dispersion
corrections, we have also optimized several minima of [1a]+
and [1b]+ at B3LYP-D3/def2-SVP level (Tables S2 and S3a).
Interestingly, while the B3LYP results for [1b]+ indeed are
close to a fully delocalized situation, clear symmetry breaking is
still found for [1a]+ even at this computational level known to
be biased toward delocalized situations. This provides addi-
tional support to a Class II description, in particular for the
somewhat more weakly coupled [1a]+. Examination of the
optimized structures of both [1a]+ and [1b]+ at the more
suitable (even if possibly somewhat too localized) BLYP35-D3
level of theory reveals significant structural differences at the
iron centers, with the average Fe(2)−P distances ca. 0.05 Å
longer than those at Fe(1) (Table S2). There is a concomitant
decrease in Fe(2)−C(δ) distances relative to Fe(1)−C(α) of
ca. 0.015 Å ([1a]+) and ca. 0.03 Å ([1b]+). This points clearly
to localized oxidation at Fe(2). As a result of this polarized
E
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
structure, vibrational bands arising from both “symmetric” and
“asymmetric” atomic displacements in the CCCC chain
have significant oscillator strength or intensity in all of the
minima explored, giving rise to bands that fall between 1969−
1987 cm−1 and 1882−1891 cm−1 ([1a]+) or between 1952−
1971 cm−1 and 1875− 1883 cm−1 ([1b]+). The calculations not
only reproduce the frequencies of the experimentally observed
two-band pattern for [1a,b]+ (Figure 2) but also given the
greater spread of vibrational frequencies for the symmetric
vibrational mode as a function of conformation (Table 1)
reproduce the shape of the bands and the general appearance of
the spectra (Figure 2). However, the calculations do not
reproduce the relative intensities of the ν(CCCC) bands
observed experimentally due to the overestimation of charge-
localization by BLYP35-D3.
While the localized BLYP35-D3 structures (Table S2)
satisfactorily explain the appearance of the IR spectra, the S2
expectation values (ca. 0.82 compared to the nominal 0.75 for a
doublet state) indicate significant spin contamination, and
analysis of the spin-density population shows excess spin
polarization (in part with appreciably negative spin densities)
along the C4-bridge. This contrasts with analogous ruthenium
complexes [2a,b], where little spin contamination is observed at
the same computational level (Table 2). In order to provide a
better description of the spin-density distribution and in
particular better electronic excitation spectra in TDDFT
calculations (see below), single-point calculations using the
local hybrid functional Lh-SsirPW92-D344 were performed
(with D3-type dispersion corrections), using a recent efficient
implementation.45,46
Figure 4. Frontier molecular orbital plots (isosurface: 0.02 au) and
spin-density plots (isosurface: 0.001 au) for [1a]+ (Ω = 55°) and
[1b]+ (Ω = 36°) using Lh-SsirPW92-D3/def2-SVP (red color in the
spin density marks negative values). Plots of related orbitals and spin-
density distributions for other minima are given in the Supporting
Information.
SOMO composition but also some spin polarization effects (cf.
negative spin density in red color). Similar molecular orbital
and spin-density distributions are found for the other
conformers, while it is apparent that the structures become
even more localized for the more “perpendicular” conformers
(in particular with Ω ≈ 90°; Tables S3 and S4, Figures S8 and
S9). In contrast to the delocalized picture established for the
ruthenium complexes (Tables S3 and S4, Figures S8 and
S9),14,16 the spin densities (and the structures) clearly point to
a partially localized picture for iron complexes [1a]+ and [1b]+.
The more localized description of iron complexes [1a]+ and
[1b]+ furthermore leads to a relatively small contribution of the
C4 bridging ligand of about 25−30% to the overall spin density
(adding up both positive and negative atomic spin
populations), while the C4 fragment supports almost 50% of
the spin density for ruthenium systems [2a]+ and [2b]+. It
should be noted in passing that the Lh-SsirPW92-D3 local
hybrid gives a somewhat less localized and less spin-polarized
distribution for the iron complexes than the BLYP35-D3 global
hybrid (Table S3, Figure S8), even when using the BLYP35-D3
structures. The smaller valence-shell spin polarization over the
bridge is directly related to the somewhat lower S2 expectation
value.
Related local hybrids, which in contrast to BLYP35 or
B3LYP exhibit position-dependent rather than constant exact-
exchange admixture, have recently been shown to provide a
good description of both localized and delocalized mixed-
valence systems,47 while reducing the spin contamination
compared to “global” hybrids with high exact exchange, such as
BLYP35. The structures were not optimized with local hybrids,
as the higher computational demand of the current gradient
implementation for local hybrids was not warranted by the
expected small structural improvements.48
Table 2 compares the energies and S2 expectation values of
BLYP35-D3 and Lh-SsirPW92-D3 single-point calculations for
all four complexes studied. Use of the local hybrid leaves the
global minima generally unchanged. The most notable change
at the Lh-SsirPW92-D3 level is the relative destabilization of
the most localized/polarized conformers near Ω = 90°, making
them the least stable ones for both [1a]+ and [1b]+ (Table 2).
We also note a smaller conformational energy spread for [1a]+
compared to that of [1b]+ and particularly compared to the two
ruthenium complexes [2a,b]+.
The spin-density distribution in the monocationic iron
complexes (Lh-SsirPW92-D3/def2-SVP level; Table S3, Figure
S8) largely tracks the composition of the frontier orbitals
(Table S4, Figure S9) and supports the description of [1a]+ and
[1b]+ in terms of a localized electronic structure (i.e., as a
Robin/Day Class II mixed-valence complex; Figure 4). That is,
for the global minima the SOMO is distributed unevenly over
the two metal centers and the bridge (SOMO Fe(1)/C4/Fe(2):
[1a]+, Ω = 55°, 7/25/57%; [1b]+, Ω = 36°, 10/29/51%). The
HOMO is similarly polarized, but localized on the other metal
center (HOMO Fe(1)/C4/Fe(2): [1a]+, Ω = 55°, 49/31/9%;
[1b]+ Ω = 36°, 50/25/15%). Plots of these orbitals and total
spin density (Figure 4) show that the latter reflects not only the
The asymmetry and associated dipole moment computed
across the long axis of polarized [1a]+ and [1b]+ readily explain
the intensity of both the “symmetric” and “asymmetric” ν(C
CCC) stretches observed in their IR spectra, indicating that
the electronic structure remains localized on the IR time scale
(ca. 1013 s−1).49,50 In contrast, for the more delocalized
16
ruthenium and rhenium complexes [2a]+, [2b]+ and
[{Re(PPh3)(NO)Cp*}2(μ-CCCC)]+,6 the symmetric
stretch only appears as an exceptionally weak band, which
only gains intensity due to the small polarization of the most
perpendicular conformers. Consequently, only the ν(CCC
C) stretch corresponding to the asymmetric atom displacement
is observed to have any appreciable intensity, and the IR spectra
are dominated by a single, strong ν(CCCC) band.
A further characteristic feature of mixed-valence complexes is
the low-energy “intervalence charge transfer” (IVCT) band,
which is often observed in the NIR to IR region. In the case of
[2a]+ and [2b]+, the NIR band envelope has been assigned on
F
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Figure 5. Spectroelectrochemically generated NIR spectra of [1a,b]+ and [2a,b]+ in CH2Cl2/0.1 M NBu4[PF6] at room temperature.
the basis of a detailed computational (TDDFT with BLYP35-
D3), structural, and spectroscopic study, and was shown to
arise from a series of HOMO-1 to SOMO transitions with
significant π−π* character involving orbitals delocalized over
the 6-atom Ru−C4−Ru chain (Table S5).16 In the conformers
with the most perpendicular mutual arrangements of the half-
sandwich fragments, these transitions are blue-shifted, account-
ing for the high-energy shoulder, and gain a degree of MLCT
character (Figure 5). In addition, a weak electronic transition in
the IR region is also observed. It arises from π−π* type
transitions (HOMO-SOMO) between orbitals that are
essentially orthogonal π-faces of the butadiyne bridge and
delocalized over the Ru−C4−Ru chains. This transition might
be approximated as a dd transition if the contributions from the
carbon atoms are ignored. Further single-point TDDFT
calculations performed here with the local hybrid functional
(Lh-SsirPW92-D3) at these previously identified structures give
essentially the same description of the electronic spectra
(Tables S5 and S6).
environment within the cell (Figure 5), are dominated by a
broad and relatively intense band envelope (ε = 12 000 ([1a]+);
22 000 ([1b]+) M−1 cm−1) that extends between ca. 12 000−
5000 cm−1, with a higher-energy shoulder that is most
pronounced for [1a]+. This absorption band envelope has
been assigned as the “Class III” IVCT (or charge resonance)
transition in previous studies of [1a]+.2−5
While use of the local hybrid thus does not alter the picture
for the ruthenium complexes relative to BLYP35-D3 data, the
BLYP35-D3 excitation spectra for [1a]+ and [1b]+ do not
reproduce the experimental spectra at all well (Figure S10).
Using the Lh-SsirPW92-D3 local hybrid instead leads to
dramatically improved agreement with experiment (Table S6,
Figure S10), which is the main reason for extending our
previous computational protocol to local hybrids in the case of
the iron complexes.41,42,51 Results for the range of conforma-
tional minima (see above) of the iron complexes are given in
Table 3. The excitations include a very low-energy electronic
transition, E1, in the IR region ([1a]+ E1 = 2416−2993 cm−1);
[1b]+ E1 = 2925−3689 cm−1) and well beyond the range of the
NIR spectrometer used in the original report.3 While
overlapped by vibrational features, the corresponding band
3,5
The NIR region of [1a]+ and [1b]+, both of which have
been recorded here using spectroelectrochemical techniques,
taking care to ensure a rigorously dry and anaerobic
G
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Table 3. TDDFT (Lh-SsirPW92-D3) Electronic Excitation Energies with Corresponding Transition Dipole Moments of
Different Conformers of [1a,b]+, [2a,b]+, and Corresponding Experimental Data
Ω (°)
E1 (cm−1
)
μ (D)
E2 (cm−1
)
μ (D)
E3 (cm−1
)
μ (D)
E4 (cm−1
)
μ (D)
E5 (cm−1
)
μ (D)
E6 (cm−1
)
μ (D)
55
2900
2416
2667
3168
2380
3689
3384
2925
3298
3689
2640
2545
1979
1177
805
0.5
0.4
0.9
1.2
6043
5656
1.8
0.8
0.2
0.8
11234
12405
12084
10325
9320
5.4
4.2
4.7
6.3
13255
12636
13226
10857
1.5
2.9
2.4
2.8
16266
15882
13744
13546
1.2
1.2
0.2
0.5
94
[1a]+
110
144
5819
16048
16403
0.9
0.8
6416
a
exp
7766
36
1.1
1.3
0.5
1.7
1.7
7017
2.0
2.3
0.4
1.4
0.9
9584
7.0
6.0
4.6
7.0
7.7
13794
11013
12868
13326
13676
0.9
2.9
3.4
1.0
0.6
15613
13434
16327
15492
14367
0.1
1.4
0.8
0.7
0.2
16683
16122
17109
16267
15363
0.8
1.1
1.4
1.1
0.2
51
6807
10318
12180
9635
86
6255
[1b]+
135
147
6769
7174
9310
a
exp
8332
10800
12377
12061
11531
11446
12134
12529
12420
13771
13842
13101
13948
14182
14599
13190
36
0.9
1.3
1.3
1.4
1.2
0.6
11912
11778
11178
11084
11911
12211
10340
13073
12958
12693
13139
13500
13376
11140
2.5
0.3
4.2
4.3
2.9
1.4
6.0
5.7
3.2
2.6
4.9
6.5
13702
13653
14193
14270
13952
13525
5.1
5.8
5.9
6.1
5.9
5.0
51
69
[2a]+
105
132
153
2137
2724
2230
2765
2519
836
a
exp
33
1.1
1.3
1.6
1.3
0.9
0.2
7.9
7.1
5.2
7.0
8.0
8.4
0.6
1.5
2.7
1.7
0.5
1.1
14761
14865
14570
14877
14961
1.5
3.0
5.2
3.2
1.1
51
98
[2b]+
127
150
174
2256
2779
3375
2590
a
exp
a
Apparent peak maxima (literature data for [2a]+ and [2b]+ quoted).16
can be readily observed in the IR spectra of [1a]+ (∼2500
cm−1) and [1b]+ (∼2750 cm−1) (Figures 6 and 7).
([1a]+) and between 6255 and 7174 cm−1 ([1b]+). This E2
transition arises from HOMO-4 and HOMO-5 to SOMO
excitations, admixed with smaller contributions from the
HOMO to SOMO excitation (the HOMO-5 contribution
becomes more dominant for the more perpendicular isomers).
On the basis of the composition of the HOMO-4 (Ω = 110°,
[1a]+, Fe(1)/C4/Fe(2): 40/11/8; Ω = 86°, [1b]+: 33/19/24)
and HOMO-5 (Ω = 110°, [1a]+, Fe(1)/C4/Fe(2): 7/11/47; Ω
= 86°, [1b]+: 13/10/52), the low-intensity E2 transition can be
described as arising from an MLCT excitation, although the
relatively low oscillator strength makes the definitive assign-
ment of features of this transition in the experimental spectrum
uncertain.
The E3 transitions provided by the TDDFT calculations have
much greater oscillator strength than E1 and E2 and fall between
10 325 and 12 405 cm−1 ([1a]+) and between 9310 and 12 180
cm−1 ([1b]+) across the range of conformations identified. As
with the other electronic transitions, the composition of E3 is
modestly sensitive to the conformation, decreasing in intensity
but blue-shifting in the case of the more perpendicular
conformers. These E3 transitions from the most perpendicular
conformers are in part responsible for the shoulder observed in
the experimental spectra (Figure 5). The E3 transitions of [1b]+
cover a narrower range than those of [1a]+, neatly accounting
for the narrower NIR absorption envelope observed in the Cp
derivative and the more significant shoulder seen with the Cp*
derivative. The E3 transitions involve HOMO to SOMO and
HOMO-1 to SOMO excitations and much smaller contribu-
tions from the HOMO-2 to SOMO excitation. The HOMO to
SOMO component is most significant for the most cis- and
trans-like conformers, while the HOMO-1 to SOMO
contribution dominates for perpendicular conformers and is
The TDDFT calculations attribute E1 to the HOMO-SOMO
excitation, admixed with HOMO-1 to SOMO and HOMO-3 to
SOMO character (the precise composition varies for each
conformer; Tables 3 and S6). The HOMO-1 to SOMO
excitation is a more important contribution for the more cis-
and trans-like conformers, while HOMO-3 to SOMO is more
significant in the more perpendicular conformers. The E1
transition red-shifts and decreases in intensity in the more
perpendicular conformers. Given the differences in the
localization of the HOMO, HOMO-1 and HOMO-3, all of
which feature significant Fe(1) character, and the SOMO,
which has appreciable Fe(2) character, these low-energy IR
bands for [1a,b]+ are well-described as part of the IVCT
transitions associated with Class II (valence-localized) mixed-
valence complexes based on pseudo-octahedral d5−d6 metal
fragments.52 Such descriptions should not, however, overlook
the important contributions of the C4 chain to the frontier
orbitals. The SOMO has ca. 25 or 30% C4 character in all
conformers of [1a]+ and [1b]+, respectively (even when
averaging the negative and positive regions created by spin
polarization, see above). The HOMO-1 and HOMO-3 feature
opposite trends in their carbon character, with the carbon
character decreasing in the more perpendicular conformers in
HOMO-1 and rising in the case of the HOMO-3. In terms of
the traditional two-state model of mixed-valence complexes, E1
might be described as the true IVCT transition.
The TDDFT calculations (Lh-SsirPW92-D3//BLYP35-D3)
provide a second transition of relatively low oscillator strength
(E2) that is also red-shifted and decreases in intensity for the
perpendicular conformers, falling between 5656 and 6416 cm−1
H
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Figure 6. Reversible spectral changes observed upon spectroelec-
trochemical oxidation of [1a,b] → [1a,b]+ in the 1000−10 000 cm−1
region.
Figure 7. Low-intensity Gaussian-shaped bands observed in the
∼2500 cm−1 region of the spectra of [1a]+ (upper) and [1b]+ (lower).
more prevalent in the transitions that give rise to the higher-
energy shoulder. HOMO-1 has a polarization similar to that of
HOMO; therefore, E3 also exhibits significant IVCT character.
The NIR envelope also contains transition E4, which also has
dominant HOMO-1 to SOMO character and occurs at
10 857−13 255 cm−1 ([1a]+) and at 11 013−13 794 cm−1
([1b]+), depending on conformation. It is also worth noting
that the potential energy surface of [1a]+ appears to be very
shallow in the region around Ω = 144°. When adding a
sampling point in that region (Ω = 144°), the additional
transitions result in the computed shoulder on the band around
10 000−12 000 cm−1 becoming more prominent, leading to
better agreement with the experimental spectrum (Figure S11).
While it is therefore likely that not every minimum has been
identified, the consolidated body of results strongly support the
use of the conformational distribution sampling a number of
close lying minima rather than a single point calculation to
achieve greater accuracy in the description of the electronic
absorption spectra. The solvatochromic character of this band
envelope is not dramatic, varying over only ca. 160 cm−1 from
CH2Cl2 to NCMe;3 this may be due to the effective screening
of the Fe2C4 chain from the solvent by the ancillary ligand
sphere and is still significantly greater than that of [2a]+.14
The highest-energy optical absorption bands observed in the
experimental spectra of [1a]+ and [1b]+ at 15 000−17 000
cm−1 arise from E5 and E6 which have dominantly HOMO-2 to
SOMO character in each case across the range of conformers.
HOMO-2 is largely associated with Fe(1); therefore, this
highest-energy band also has a degree of metal-to-metal charge
transfer character. Multiple transitions with IVCT character are
not uncommon for localized d5/d6 dinuclear mixed-valence
complexes,52 and Figure 8 shows the good match of the
computed spectra (obtained when convoluting the stick spectra
of the various conformers at Lh-SsirPW92-D3//BLYP35-D3
level with Gaussian functions of half-width 800 cm−1) with the
spectroelectrochemically determined experimental spectra.
Finally, it is interesting to consider the underlying reasons for
the distinct electronic characters of more localized iron
I
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
CONCLUSIONS
■
The present organometallic C4-bridged mixed-valence iron and
ruthenium complexes, [1a,b]+ and [2a,b]+, respectively, have
been prototype systems for the study of mixed-valency, electron
transfer, and in general all-carbon bridging ligands for several
decades. The present finding that diiron complexes [1a,b]+ are
on the localized Class II side of the mixed-valence spectrum, at
least on the time scale of the IR experiment (10−13 s), which
puts them into marked contrast to the delocalized Class III
ruthenium complexes [2a,b]+, while thus far much more
similarity between the two types had been assumed. This
interesting electronic-structure counterpoint highlights the
differences in metal−ligand overlap for 3d and 4d systems
and indicates that in the case of “all-carbon” or “carbon-rich”
bridged bimetallic complexes first row metals are more likely to
result in “localized” (Class II) behavior while second row
metals likely give more delocalized structures. Of course, the
question of “localized” or “delocalized” character is intricately
related to the time-scale of observation, and on the longer time
scale of ESR and Mossbauer spectroscopies, [1a]+ appears to be
more fully delocalized. The effective electron transfer rate
between the two iron centers can be estimated as between 10−9
and 10−13 s.
The localized character of the two diiron systems manifests
itself in the observation of symmetric and asymmetric ν(C
CCC) stretches with significant IR intensity. This contrasts
with barely detectable symmetric stretches for the delocalized
ruthenium and closely related rhenium examples, where the
symmetric stretch only gains some intensity due to the larger
symmetry breaking in the most perpendicular conformers. A
comprehensive DFT investigation has identified a range of
structural minima, which differ in the relative orientation of the
half-sandwich fragments across the C4 ligand but have structural
features consistent with oxidation at only one iron center and
localized on the IR time scale. Harmonic vibrational frequency
analyses for these conformers account for the appearance of the
experimental IR spectra. Moreover, the calculations indeed
show [1a,b]+ to be localized Class II and [2a,b]+ to be
delocalized Class III.
Electronic excitation spectra in the NIR (and IR) region also
reflect the rich conformational landscape, and the computations
help explain some intriguing features of the spectra, such as the
identification of a new, low-energy IVCT band in the IR region
(for both the iron and ruthenium systems). Further electronic
absorptions in the NIR region arise from overlapping IVCT-like
transitions in each of the thermally accessible conformers. For
the study of these electronic excitations, we had to extend our
BLYP35-D3-based computational protocol for the treatment of
mixed-valence systems, as while the spectra for the diruthenium
complexes are well-reproduced with BLYP35-D3, this is not the
case for the mixed-valence diiron systems, likely due to the
onset of significant spin contamination. Here the use of a novel
type of exchange-correlation functionals, so-called local hybrids
with position-dependent exact-exchange admixture, provided
the necessary improvements. Thus, whilst the electronic
structure and spectra are essentially unchanged compared to
BLP35-D3 for ruthenium complexes, the calculated electronic
spectra of the iron complexes were decisively improved by the
local hybrid, allowing their detailed interpretation.
Figure 8. Spectroelectrochemically generated NIR spectra of [1a,b]+
(green, dotted lines) and computed spectra, convoluting a simple
average of the stick spectra of the various conformers (black, solid line,
with stick spectra).
complexes [1a]+, [1b]+ and more extensively delocalized
ruthenium complexes [2a]+, [2b]+. The degree of electronic
interaction between the two half-sandwich redox centers;
hence, the degree of (de)localization is a result of the overlap of
the valence orbitals of the metal centers Fe: 3d, Ru: 4d) and the
bridging ligand (C4: π-system). Due to their larger spatial
extension, the 4d orbitals of ruthenium overlap better with the
π-system of the butadiyndiyl-bridge than do the 3d orbitals of
iron which are exceptionally compact due to the lack of a radial
node.53 These differences in atomic orbital structure contribute
to more delocalization of charge in case of the largely Class III
ruthenium systems while the iron complexes exhibit an
appreciably more localized Class II behavior. In addition, the
role that ancillary ligands may play through steric and
dispersion effects on the conformational distribution and
hence underlying electronic properties is also worthy of further
investigation; complex [1c]+ represents a useful target through
which to explore these ancillary ligand effects, which will be the
subject of later studies.
The presence of multiple conformers in all but the most
symmetric of “mixed-valence” systems, all of which offer subtly
different electronic transitions, making application of band-
J
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
(br, 4H, CH2, dppe), 4.10 (br, 2H, CCH), 4.86 (s, 10H, Cp), 6.91,
7.17, 7.28, 7.38, (m, 40H, Ph, dppe). 13C{1H} NMR (CD2Cl2): 358.1
(t, JC−P 35 Hz), Cα; 135.9, m, 134.3, m, Phi; 132.5, m, 131.1, m, Pho;
131.4, s, 131.2, s, Php; 129.2, m, 129.1, m, Phm; 109.6, s, (CCH);
89.2, s, Cp; 28.5, m, CH2 (dppe). 31P{1H} NMR (CD2Cl2): δ 95.7.
FT-IR (ATR) ν 1636(w), 1603(m) cm−1 (CC). MALDI-MS (m/
z): 1086 [M − 2H]+. for C66H60Fe2P6F12·2CH2Cl2: C, 52.7, H, 4.1.
Found: C, 53.4; H, 4.0.
shape analysis within the framework of Marcus−Hush theory
fraught with difficulty. The apparent band envelope maxima
and molar extinction coefficients of the absorption spectra arise
from these overlapping transitions from different conformers
and cannot necessarily be attributed to any single conformer.
Uncertainties in the precise composition of the equilibrated
conformational mixture mean that accurate deconvolution and
attribution of components of the band envelope to any
individual conformer or electronic transition is not possible.
Detailed computational analyses of spin-density distributions
and of the composition of the frontier molecular orbitals
nevertheless allow the electronic coupling between the redox
centers and the overall electronic structure as well as the
spectroscopy of such systems to be understood, provided
appropriate computational approaches are applied.
[{Cp(dppe)Fe}(μ-CCCC){Fe(dppe)Cp}] [1b]. A mixture of
[{Fe(dppe)Cp}2(μ-CCH−CHC)][PF6]2 (0.77 g, 0.56 mmol)
and potassium tert-butoxide (0.13 g, 1.19 mmol) in tetrahydrofuran
(20 mL) was stirred at 0 °C for 15 min to give an orange-brown
solution which was then evaporated to dryness. The reaction residue
was extracted with toluene, and the extract was filtered through celite,
evaporated to dryness, and the residue recrystallized from CH2Cl2/n-
hexane to afford the product as an orange-brown solid in 16% yield
1
(0.10 g, 0.09 mmol). H NMR (CD2Cl2/CoCp2): δ 1.89 (br, 4H,
CH2, dppe), 2.16 (br, 4H, CH2, dppe), 3.99 (s, 10H, Cp), 7.00, 7.11,
7.24, 7.63, (m, 40H, Ph, dppe). 13C{1H} NMR (CD2Cl2/CoCp2):
142.9, m, 139.2, m, Phi; 133.7, m, 131.9, m, Pho; 128.9, s, 128.5, s, Php;
127.8, m, 127.4, m, Phm; 110.2, s, Cβ; 88.9 (t, JC−P 44 Hz), Cα; 78.8, s,
Cp; 28.0, m, CH2 (dppe). 31P{1H} NMR (CD2Cl2/CoCp2): δ 105.0.
FT-IR (CH2Cl2) ν 1955 (ATR) ν 1925 cm−1 (CC). ES(+) MS (m/
z): 1086 [M]+. Anal. Calcd. (%) for C66H58Fe2P4: C, 72.9, H, 4.3.
Found, C, 72.8, H, 5.6. A single crystal of [1b] suitable for X-ray
diffraction studies was obtained by slow diffusion of n-pentane into a
dichloromethane solution of [1b] at ambient temperature (ca. 22 °C)
giving the bis(CH2Cl2) solvate. C68H62Cl4Fe2P4, M = 1256.56, dark-
cherry prism, 0.36 × 0.26 × 0.20 mm3, orthorhombic, space group
Pbca (No. 61), a = 18.1113(1) Å, b = 20.7731(2) Å, c = 31.5098(3) Å,
V = 11854.86(17) Å3, Z = 8, Dc = 1.408 g/cm3, μ = 0.820 mm−1. F000
= 5200, Mo Kα radiation, λ = 0.71073 Å, T = 100(2) K, 2θmax = 75.1°,
38 8734 reflections collected, 30 686 unique (Rint = 0.0919). Final
GooF = 1.000, R1 = 0.0560, wR2 = 0.1278, R indices based on 21 585
reflections with I > 2σ(I) (refinement on F2), |Δρ|max = 2.1(1) e Å−3,
703 parameters, 0 restraints. Lp and absorption corrections applied.
[{Cp(dppe)Fe}(μ-CCCC){Fe(dppe)Cp}][PF6]2 ([1b][PF6]2).
Ferrocenium hexafluorophosphate (0.28 g, 0.85 mmol) was added to
a solution of 1b (0.50 g, 0.46 mmol) in dichloromethane (20 mL) at 0
°C, and the reaction mixture was stirred at this temperature for 1 h
resulting in a color change from orange to blue-green. The reaction
mixture was filtered through Celite and treated with diethyl ether to
precipitate the product in 70% yield as a deep green solid (2·CH2Cl2
solvate; 0.50 g, 0.32 mmol). FT-IR (ATR) ν 1808 cm−1 (CC).
MALDI-MS (m/z) [M]+ 1086. Anal. Calcd: (%) for C66H58Fe2P6F12·
2CH2Cl2: C, 52.8, H, 4.0. Found: C, 53.4, H, 3.9.
EXPERIMENTAL DETAILS
■
FeI(CO)2Cp,54 FeCl(dppe)Cp,55 and 1a3 were synthesized according
to existing literature procedures. The preparation, purification, and
reactions of the complexes described were carried out under dry
nitrogen. All solvents were dried by standard methods, distilled, and
deoxygenated before use. NMR spectra were recorded on a Bruker
Avance III HD (500 MHz 1H, 125 MHz 13C{1H}, 202 MHz 31P{1H})
spectrometer and referenced to internal solvent references (1H,
13C{1H})56 or external reference 85% H3PO4 (31P{1H} δ = 0.0).
Infrared spectra (solid state) were obtained on a Nicolet iS5 FT-IR or
an Agilent Technologies Cary 630 spectrometer fitted with an ATR
attachment. Electrospray mass spectra were obtained on a Waters
SQD2 instrument, and MALDI mass spectra were recorded using a
Shimadzu Axima Confidence spectrometer. Microanalyses were
conducted by the staff of the Microanalytical Service of the School
of Chemistry, University of Manchester. Cyclic voltammetry was
carried out as previously reported using the decamethylferrocene/
decamethylferrocinium (FeCp*2/[FeCp*2]+) couple as an internal
reference for potential measurements (as the usual ferrocene reference
couple overlapped with a wave in the CV of 1b) such that the couple
falls at −0.55 V (CH2Cl2/NBu4PF6) relative to external FeCp2/
[FeCp2]+ at 0.00 V.57 Spectroelectrochemical measurements were
conducted as previously described.16
Fe(CCH)(dppe)Cp. Ethynyltrimethylsilane (443 mg, 4.51
mmol) was added to a solution of FeCl(dppe)Cp (500 mg, 903
μmol) and ammonium hexafluorophosphate (294 mg, 1.81 mmol) in
methanol (10 mL); the reaction mixture was heated at reflux for 1 h
and then allowed to cool to ambient temperature. Subsequently,
sodium (104 mg, 4.51 mmol) was added to the stirred orange
suspension resulting in a mildly exothermic reaction. This mixture was
allowed to stir for a further 1 h at ambient temperature and filtered to
afford Fe(CCH)(dppe)Cp in 73% yield (358 mg, 660 μmol) as an
red/orange powder which was washed with methanol (20 mL) and
dried in air. All analytical data were in agreement with those reported
in the literature.58 1H NMR (500.1 MHz CDCl3): δ = 1.31 (s, 1 H,
CCH), 2.26 (m, 2 H, dppe), 2.72 (m, 2 H, dppe), 4.21 (s, 5 H,
C5H5), 7.15 (m, 4 H, o-C6H5, Fe(dppe)Cp), 7.24 (m, 4 H, m-C6H5,
Fe(dppe)Cp), 7.30 (m, 4 H, m-C6H5, Fe(dppe)Cp), 7.38 (m, 4 H, p-
C6H5, Fe(dppe)Cp), 7.90 (m, 4 H, o-C6H5, Fe(dppe)Cp). 31P NMR
(202.4 MHz, CDCl3): δ = 106.3 (s). FT-IR (CH2Cl2) ν 1917 (CC),
3271 cm−1 (C−H).
[{Cp(dppe)Fe}(μ-CCCC){Fe(dppe)Cp}][PF6] ([1b]PF6). A
sample of [1b][PF6]2 (0.50 g, 0.36 mmol) was dissolved in
dichloromethane (25 mL) at 0 °C, 1b (0.39 g, 0.36 mmol), and the
reaction mixture was stirred at 0 °C for 1 h. A black-green solution
formed which was filtered through celite. Diethyl ether added to the
filtrate resulting in formation of a precipitate which was isolated by
filtration, washed with diethyl ether, and dried in vacuo affording the
product in 53% yield as a black-green precipitate (CH2Cl2 solvate; 0.50
g, 0.38 mmol). FT-IR (CH2Cl2) ν 1877, 1980 cm−1 (CC). MALDI-
MS (m/z) [M]+ 1086. Anal. Calcd: (%) for C66H58Fe2P5F6·CH2Cl2:
C, 61.1%; H, 4.6%. Found, C, 61.0% H, 4.3%.
[Fe(CO)(dppe)Cp][BPh4]. A mixture of [FeI(CO)2Cp] (0.43 g,
1.41 mmol), diphenylphosphinoethane (0.56 g, 1.41 mmol) and
sodium tetraphenylborate (0.96 g, 2.81 mmol) in toluene (50 mL) was
heated at reflux with stirring for 18 h, resulting in the formation of a
yellow-brown precipitate. The reaction mixture was cooled, and the
precipitate was collected. Recrystallization from dichloromethane/
diethyl ether gave the product in 61% yield as a yellow solid (0.74 g,
[{Fe(dppe)Cp}2(μ-CCH-CHC)][PF6]2. Ferrocenium hexafluor-
ophosphate (1.02 g, 3.09 mmol) was added to a solution of [Fe(C
CH)(dppe)Cp] (1.68 g, 3.09 mmol) in dichloromethane (12 mL) at
−78 °C; the reaction mixture was stirred at this temperature for 3 h.
After warming to −20 °C, diethyl ether was added to be just sufficient
to induce precipitation of the required product as a green-brown solid;
the remaining orange-brown solution contains the more soluble
reaction byproduct, [Fe(CCH2)(dppe)Cp][PF6].59 Recrystalliza-
tion of the crude product from CH2Cl2/diethyl ether afforded the
product in 83% yield as a green-brown solid (2·CH2Cl2 solvate, 1.98 g,
1
0.85 mmol). H NMR (CDCl3): δ 2.04, 2.19 (br, 4H, CH2, dppe),
4.40 (s, 5H, Cp), 6.71, 6.84, 6.95, 7.18, 7.32, 7.43 (m, 40H, Ph, dppe).
31P{1H} NMR (CDCl3): δ 92.9. FT-IR (CH2Cl2) ν 1979 (ATR) ν
1970 cm−1 (CO). MALDI/MS (m/z): 547 [M]+. Anal. Calcd: (%)
for C56H49FeP2OB: C, 77.6%; H, 5.7%. Found, C, 77.2% H, 5.6%.
1
1.28 mmol). H NMR (CD2Cl2): δ 2.56 (br, 4H, CH2, dppe), 2.78
K
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
COMPUTATIONAL DETAILS
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.organo-
met.8b00099.
■
■
S
All structure optimizations and property calculations used the
TURBOMOLE program (versions 6.460 and 7.1,61 locally modified
by the Berlin group). The COSMO continuum solvent model for
dichloromethane (DCM, ε = 8.93)62 was generally applied, using
nonequilibrium solvation in TDDFT calculations.62,63 Due to the
involvement of the solvent model, vibrational frequency calculations
were done by numerical differentiation of analytical gradients using
TURBOMOLE’s NumForce script (SCF convergence 10−8 Eh). The
resulting harmonic vibrational frequencies were scaled by an empirical
factor of 0.895.43
All structure optimizations and vibrational frequency analyses were
carried out with the BLYP35 functional,41 a global hybrid based on
B88 exchange and LYP correlation (BLYP functional)64,65 using 35%
exact-exchange admixture. BLYP35 has been found to perform well for
mixed-valence (MV) systems near the Class II/III borderline in the
Robin−Day scheme.14,41,42,51,66 Additional structure optimizations and
vibrational frequency calculations were performed with the well-known
global hybrid functional B3LYP,65,67−69 using the full ligand sphere,
solvent model, and dispersion correction to provide comparison with
the results from earlier gas-phase calculations on truncated models of
[1b]+ (substituting phenyl moieties of the dppe ligands with hydrogen
atoms).24
Plots of the molecular structure of 1b and summary of
crystal data, the cyclic voltammogram of 1b, the IR
spectra of [{Fe(dppe)Cp*}2(μ-CC−CC)]n+
([1a]n+) (n = 0−2), tabulated ν(CC) data (from
CH2Cl2 solutions) from selected complexes [{LmM}(μ-
CCCC){MLm}]n+, comparing the solid and sol-
ution state spectra of [1a]+ and [1b]+, the relative
energies (BLYP35-D3/def2-SVP) of various conformers
of [1b]+, [2a]+ and [2b]+, and experimental NIR spectra
of [1a]+ and [1b]+ overlaid with convoluted Gaussian
broadened (line width: 800 cm−1) stick spectra from Lh-
SsirPW92-D3 and BLYP35-D3 calculated spectra;
computed structural properties of all minima at
BLYP35-D3 level and those for [1a]+ and [1b]+ at
B3LYP-D3 level; Mulliken spin densities and associated
plots; molecular orbital populations and associated plots,
and excitation energies and tranistion dipole moments;
transition character of electronic excitations of different
conformers at Lh-SsirPW92-D3 level of theory; relative
energies (kJ mol−1), total spin expectation values, and
computed harmonic vibrational frequencies for different
conformers of [1a]+ and [1b]+ at B3LYP-D3/def2-SVP
level of theory (PDF)
Due to the observation of appreciable spin contamination and poor
reproduction of the excitation energies for the present iron complexes
using BLYP35, additional single-point calculations of ground-state
energies and TDDFT calculations of excitation spectra used the Lh-
SsirPW92 local hybrid functional with position-dependent exact-
exchange admixture and a partially self-interaction-corrected correla-
tion part.44 The local hybrid functional is constructed according to
EXLh
=
a(r) εexact(r) dr + [1 − a(r)] εDFT(r) dr + ECDFT
(XYZ)
∫
∫
X
X
Accession Codes
where the constant admixture parameter (global hybrid functional, e.g.,
0.35 for BLYP35, 0.2 for B3LYP) has been replaced by a real-space
dependent local mixing function a(r):
CCDC 1570320 contains the supplementary crystallographic
data for this paper. These data can be obtained free of charge
via www.ccdc.cam.ac.uk/data_request/cif, or by emailing data_
request@ccdc.cam.ac.uk, or by contacting The Cambridge
Crystallographic Data Centre, 12 Union Road, Cambridge CB2
1EZ, UK; fax: +44 1223 336033.
a(r) = 0.646τW/τ
where τW and τ are the von Weizsacker and the Kohn−Sham kinetic-
̈
energy densities, respectively.
This and closely related local hybrids have not only been found to
perform extremely well in TDDFT calculations of a wide range of
excitation classes70 but also have interesting properties regarding the
treatment of mixed-valence systems,47 thus extending our general
computational protocol. The calculations used the recent efficient
seminumerical implementation of local hybrids into TURBOMOLE
7.161 for ground-45 and TDDFT excited-state46 calculations (in the
latter case extending the available Turbomole TDDFT implementa-
tion71).
AUTHOR INFORMATION
Corresponding Authors
*Phone: +49 30 314 79682. E-mail: martin.kaupp@tu-berlin.de
*Phone: +61(8)6488-3045. E-mail: paul.low@uwa.edu.au.
ORCID
Martin Kaupp: 0000-0003-1582-2819
Paul J. Low: 0000-0003-1136-2296
Notes
■
In general, def2-SVP basis sets were used.72 Earlier calculations on
the ruthenium complexes using larger basis sets (def2-TZVP) gave
negligible changes in structure and spectroscopic parameters.16
Numerical integration grids of size m3 (grid 1 for the SCF iterations,
grid 3 for the final energy evaluation) were applied. Semiempirical
dispersion correction terms within Grimme’s DFT-D3 approach have
been added,73 using DFT-D3 parameters rs6 = 1.1225 and s8 = 0.9258
optimized for the BLYP35 functional74 and rs6 = 0.77 and s8 = 1.429
for the Lh-SsirPW92 functional.75
A relaxed scan was performed (BLYP35-D3) for computationally
less demanding compound [1b]+, evaluating the potential-energy
surface along different P−Fe···Fe−P torsional angles (Ω) in steps of
10°. Unconstrained structure optimizations of the minima obtained
during the scan were then done. Similar relaxed scans for [2a,b] have
been taken from our previous work.14,16 Molecular-orbital and spin-
density isosurface plots were obtained with the MOLEKEL software
package.76 For [1a]+, seven starting structures with torsional angles
taken from [2a]+ were used to investigate its potential energy surface
(PES).
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This research was supported by the ARC (DP 140100855) and
DFG (KA1187/13-2). P.J.L. held an ARC Future Fellowship
(FT 120100073) and holds a Friedrich Wilhelm Bessel
Research Award from the Alexander von Humboldt Founda-
tion. This project was further supported by the North-German
Supercomputing Alliance (HLRN), which provided HPC
resources for numerical frequency calculations. We thank the
Government of Libya for a research studentship (to S.E.-T.)
and the DAAD/Go8 for a travel grant which has enhanced the
collaboration between Perth and Berlin. We are also grateful to
Mr. Damien Bainbridge for the construction of spectroscopy
cell holders, which greatly facilitated the spectroelectrochemical
measurements.
L
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
White, A. H.; Zaitseva, N. N. Z. Anorg. Allg. Chem. 2013, 639, 2216−
2223.
(31) Catheline, D.; Astruc, D. Organometallics 1984, 3 (7), 1094−
1100.
REFERENCES
■
(1) Bruce, M. I.; Low, P. J. Adv. Organomet. Chem. 2004, 50, 179−
444.
(2) Le Narvor, N.; Lapinte, C. J. Chem. Soc., Chem. Commun. 1993,
(32) Riley, P. E.; Davis, R. E. Organometallics 1983, 2 (2), 286−292.
(33) Paul, F.; Toupet, L.; Roisnel, T.; Hamon, P.; Lapinte, C. C. R.
Chim. 2005, 8 (8), 1174−1185.
(34) Krejcik, M.; Danek, M.; Hartl, F. J. Electroanal. Chem. Interfacial
Electrochem. 1991, 317 (1−2), 179−187.
(35) Lapinte, C. J. Organomet. Chem. 2008, 693 (5), 793−801.
(36) Winter, R. F. Organometallics 2014, 33 (18), 4517−4536.
(37) D’Alessandro, D. M.; Keene, F. R. Dalton Trans 2004, No. 23,
3950−3954.
(38) Low, P. J.; Brown, N. J. J. Cluster Sci. 2010, 21 (3), 235−278.
(39) Renz, M.; Kess, M.; Diedenhofen, M.; Klamt, A.; Kaupp, M. J.
Chem. Theory Comput. 2012, 8 (11), 4189−4203.
(40) Renz, M.; Kaupp, M. J. Phys. Chem. A 2012, 116 (43), 10629−
10637.
No. 4, 357−359.
(3) Le Narvor, N.; Toupet, L.; Lapinte, C. J. Am. Chem. Soc. 1995,
117 (27), 7129−7138.
(4) Guillemot, M.; Toupet, L.; Lapinte, C. Organometallics 1998, 17
(10), 1928−1930.
(5) Halet, J. F.; Lapinte, C. Coord. Chem. Rev. 2013, 257 (9−10),
1584−1613.
(6) Brady, M.; Weng, W. Q.; Zhou, Y. L.; Seyler, J. W.; Amoroso, A.
J.; Arif, A. M.; Bohme, M.; Frenking, G.; Gladysz, J. A. J. Am. Chem.
Soc. 1997, 119 (4), 775−788.
(7) Jiao, H. J.; Costuas, K.; Gladysz, J. A.; Halet, J. F.; Guillemot, M.;
Toupet, L.; Paul, F.; Lapinte, C. J. Am. Chem. Soc. 2003, 125 (31),
9511−9522.
(8) Bruce, M. I.; Denisovich, L. I.; Low, P. J.; Peregudova, S. M.;
Ustynyuk, N. A. Mendeleev Commun. 1996, No. 6, 200−201.
(9) Bruce, M. I.; Low, P. J.; Costuas, K.; Halet, J. F.; Best, S. P.;
Heath, G. A. J. Am. Chem. Soc. 2000, 122 (9), 1949−1962.
(10) Bruce, M. I.; Ellis, B. G.; Low, P. J.; Skelton, B. W.; White, A. H.
Organometallics 2003, 22 (16), 3184−3198.
(11) Bruce, M. I.; Costuas, K.; Davin, T.; Ellis, B. G.; Halet, J. F.;
Lapinte, C.; Low, P. J.; Smith, M. E.; Skelton, B. W.; Toupet, L.;
White, A. H. Organometallics 2005, 24 (16), 3864−3881.
(12) Bruce, M. I.; Kramarczuk, K. A.; Skelton, B. W.; White, A. H. J.
Organomet. Chem. 2010, 695 (3), 469−473.
(41) Renz, M.; Theilacker, K.; Lambert, C.; Kaupp, M. J. Am. Chem.
Soc. 2009, 131 (44), 16292−16302.
(42) Parthey, M.; Kaupp, M. Chem. Soc. Rev. 2014, 43 (14), 5067−
5088.
(43) Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100 (41), 16502−
16513.
(44) Arbuznikov, A. V.; Kaupp, M. J. Chem. Phys. 2012, 136 (1),
014111.
(45) Bahmann, H.; Kaupp, M. J. Chem. Theory Comput. 2015, 11 (4),
1540−1548.
(46) Maier, T. M.; Bahmann, H.; Kaupp, M. J. Chem. Theory Comput.
2015, 11 (9), 4226−4237.
(13) Bruce, M. I.; Le Guennic, B.; Scoleri, N.; Zaitseva, N. N.; Halet,
J. F. Organometallics 2012, 31 (13), 4701−4706.
(14) Parthey, M.; Gluyas, J. B. G.; Schauer, P. A.; Yufit, D. S.;
Howard, J. A. K.; Kaupp, M.; Low, P. J. Chem. - Eur. J. 2013, 19 (30),
9780−9784.
(47) Kaupp, M.; Karton, A.; Bischoff, F. A. J. Chem. Theory Comput.
2016, 12 (8), 3796−3806.
(48) Klawohn, S.; Bahmann, H.; Kaupp, M. J. Chem. Theory Comput.
2016, 12 (9), 4254−4262.
(15) Gluyas, J. B. G.; Sobolev, A. N.; Moore, E. G.; Low, P. J.
Organometallics 2015, 34 (16), 3923−3926.
(49) Ito, T.; Hamaguchi, T.; Nagino, H.; Yamaguchi, T.; Kido, H.;
Zavarine, I. S.; Richmond, T.; Washington, J.; Kubiak, C. P. J. Am.
Chem. Soc. 1999, 121 (19), 4625−4632.
(16) Gluyas, J. B. G.; Guckel, S.; Kaupp, M.; Low, P. J. Chem. - Eur. J.
2016, 22 (45), 16138−16146.
(17) Venkatesan, K.; Fox, T.; Schmalle, H. W.; Berke, H.
Organometallics 2005, 24 (12), 2834−2847.
(50) Ito, T.; Hamaguchi, T.; Nagino, H.; Yamaguchi, T.; Washington,
J.; Kubiak, C. P. Science 1997, 277 (5326), 660−663.
(51) Kaupp, M.; Renz, M.; Parthey, M.; Stolte, M.; Wurthner, F.;
Lambert, C. Phys. Chem. Chem. Phys. 2011, 13 (38), 16973−16986.
(52) Demadis, K. D.; Hartshorn, C. M.; Meyer, T. J. Chem. Rev.
2001, 101 (9), 2655−2685.
(18) Venkatesan, K.; Blacque, O.; Berke, H. Dalton Trans 2007,
No. 11, 1091−1100.
(19) Fritz, T.; Schmalle, H. W.; Blacque, O.; Venkatesan, K.; Berke,
H. Z. Anorg. Allg. Chem. 2009, 635 (9−10), 1391−1401.
(20) Semenov, S. N.; Taghipourian, S. F.; Blacque, O.; Fox, T.;
Venkatesan, K.; Berke, H. J. Am. Chem. Soc. 2010, 132 (22), 7584−
7585.
(53) Kaupp, M. J. Comput. Chem. 2007, 28 (1), 320−325.
(54) El-Tarhuni, S.; Ho, M.; Kawser, M. H.; Shi, S. Q.; Whiteley, M.
W. J. Organomet. Chem. 2014, 752, 30−36.
(21) Lissel, F.; Schwarz, F.; Blacque, O.; Riel, H.; Lortscher, E.;
Venkatesan, K.; Berke, H. J. Am. Chem. Soc. 2014, 136 (41), 14560−
14569.
(22) Fitzgerald, E. C.; Brown, N. J.; Edge, R.; Helliwell, M.; Roberts,
H. N.; Tuna, F.; Beeby, A.; Collison, D.; Low, P. J.; Whiteley, M. W.
Organometallics 2012, 31 (1), 157−169.
(23) Herrmann, C.; Neugebauer, J.; Gladysz, J. A.; Reiher, M. Inorg.
Chem. 2005, 44 (18), 6174−6182.
(24) Coat, F.; Paul, F.; Lapinte, C.; Toupet, L.; Costuas, K.; Halet, J.
F. J. Organomet. Chem. 2003, 683 (2), 368−378.
(25) Beddoes, R. L.; Bitcon, C.; Ricalton, A.; Whiteley, M. W. J.
Organomet. Chem. 1989, 367 (3), C21−C24.
(26) Szafert, S.; Gladysz, J. A. Chem. Rev. 2006, 106 (11), PR1−
PR33.
(55) Gluyas, J. B. G.; Brown, N. J.; Farmer, J. D.; Low, P. J. Aust. J.
Chem. 2017, 70 (1), 113−119.
(56) Fulmer, G. R.; Miller, A. J. M.; Sherden, N. H.; Gottlieb, H. E.;
Nudelman, A.; Stoltz, B. M.; Bercaw, J. E.; Goldberg, K. I.
Organometallics 2010, 29 (9), 2176−2179.
(57) Gluyas, J. B. G.; Boden, A. J.; Eaves, S. G.; Yu, H.; Low, P. J.
Dalton Trans 2014, 43 (17), 6291−6294.
(58) Wang, Y. L.; Diallo, A. K.; Ornelas, C.; Ruiz, J.; Astruc, D. Inorg.
Chem. 2012, 51 (1), 119−127.
(59) El-Tarhuni, S.; Manhaes, L. M.; Morrill, C.; Raftery, J.;
̃
Randhawa, J. K.; Whiteley, M. W. J. Organomet. Chem. 2016, 811, 20−
25.
(60) TURBOMOLE, V6.4 2012; TURBOMOLE GmbH: Karlsruhe,
Germany, 2012. http://www.turbomole.com.
(61) TURBOMOLE, V7.1 2016; TURBOMOLE GmbH: Karlsruhe,
Germany, 2016. http://www.turbomole.com.
(62) Klamt, A.; Schuurmann, G. J. Chem. Soc., Perkin Trans. 2 1993,
No. 5, 799−805.
(63) Klamt, A. Wiley Interdiscipl Rev Comput Mol Sci 2011, 1 (5),
699−709.
(64) Becke, A. D. Phys. Rev. A: At., Mol., Opt. Phys. 1988, 38 (6),
3098−3100.
(27) Szafert, S.; Gladysz, J. A. Chem. Rev. 2003, 103 (11), 4175−
4205.
(28) Bruce, M. I.; Hall, B. C.; Kelly, B. D.; Low, P. J.; Skelton, B. W.;
White, A. H. J. Chem. Soc., Dalton Trans. 1999, No. 21, 3719−3728.
(29) Bruce, M. I.; Ellis, B. G.; Gaudio, M.; Lapinte, C.; Melino, G.;
Paul, F.; Skelton, B. W.; Smith, M. E.; Toupet, L.; White, A. H. Dalton
Trans 2004, No. 10, 1601−1609.
(30) Bruce, M. I.; Cole, M. L.; Costuas, K.; Ellis, B. G.; Kramarczuk,
K. A.; Lapinte, C.; Nicholson, B. K.; Perkins, G. J.; Skelton, B. W.;
M
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
(65) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. B: Condens. Matter
Mater. Phys. 1988, 37 (2), 785−789.
(66) Parthey, M.; Gluyas, J. B. G.; Fox, M. A.; Low, P. J.; Kaupp, M.
Chem. - Eur. J. 2014, 20 (23), 6895−6908.
(67) Becke, A. D. J. Chem. Phys. 1993, 98 (7), 5648−5652.
(68) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58 (8),
1200−1211.
(69) Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J. J.
Phys. Chem. 1994, 98 (45), 11623−11627.
(70) Maier, T. M.; Bahmann, H.; Arbuznikov, A. V.; Kaupp, M. J.
Chem. Phys. 2016, 144 (7), 074106.
(71) Bauernschmitt, R.; Haser, M.; Treutler, O.; Ahlrichs, R. Chem.
Phys. Lett. 1997, 264 (6), 573−578.
(72) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7 (18),
3297−3305.
(73) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys.
2010, 132 (15), 154104.
(74) Kaupp, M.; Guckel, S.; Renz, M.; Klawohn, S.; Theilacker, K.;
Parthey, M.; Lambert, C. J. Comput. Chem. 2016, 37 (1), 93−102.
(75) Theilacker, K.; Arbuznikov, A. V.; Kaupp, M. Mol. Phys. 2016,
114 (7−8), 1118−1127.
(76) Varetto, U. MOLEKEL 5.4; Swiss National Computing Centre:
Lugano Switzerland, 2000.
N
DOI: 10.1021/acs.organomet.8b00099
Organometallics XXXX, XXX, XXX−XXX