Solvent-Controlled CO2 Reduction by a Triphos-Iron Hydride Complex

Article abstract of DOI:10.1021/acs.organomet.8b00711

The selective reduction of CO₂ is of high interest toward future applications as a C1-building block. Therefore, metal complexes that allow for the formation of specific CO₂ reduction products under distinct reaction conditions are necessary. A detailed understanding of the CO₂ reduction pathways on a molecular level is, however, required to help in designing catalytic platforms for efficient CO₂ conversion with specific product formation. Reported herein is a unique example of a solvent-controlled reduction of CO₂ using a Triphos-based iron hydride complex. In THF, CO₂ reduction selectively leads to CO formation, whereas experiments in acetonitrile exclusively afford formate, HCOO^-. In order to explain the experimental findings, theoretical calculations of the reaction pathways were performed and further demonstrate the importance of the applied solvent for a selective reduction of CO₂.

Full text of DOI:10.1021/acs.organomet.8b00711

Article
pubs.acs.org/Organometallics
Cite This: Organometallics XXXX, XXX, XXX−XXX
Solvent-Controlled CO2 Reduction by a Triphos−Iron Hydride
Complex
Linda Iffland,†,⊥ Abhishek Khedkar,‡,§,⊥ Anette Petuker,† Max Lieb,† Florian Wittkamp,†
Maurice van Gastel,§ Michael Roemelt,*,‡,§ and Ulf-Peter Apfel*,†,∥
†
̈
̈
Anorganische Chemie I, Ruhr-Universitat Bochum, Universitatsstraße 150, 44801 Bochum, Germany
‡Lehrstuhl fu
§Max-Planck Institut fu
∥Fraunhofer UMSICHT, Osterfelder Straße 3, 46047 Oberhausen, Germany
̈
r Theoretische Chemie, Ruhr-Universitat Bochum, 44780 Bochum, Germany
̈
̈
r Kohlenforschung, Kaiser-Wilhelm-Platz 1, 45470 Mulheim an der Ruhr, Germany
̈
S
* Supporting Information
ABSTRACT: The selective reduction of CO2 is of high
interest toward future applications as a C1-building block.
Therefore, metal complexes that allow for the formation of
specific CO2 reduction products under distinct reaction
conditions are necessary. A detailed understanding of the
CO2 reduction pathways on a molecular level is, however,
required to help in designing catalytic platforms for efficient
CO2 conversion with specific product formation. Reported
herein is a unique example of a solvent-controlled reduction of
CO2 using a Triphos-based iron hydride complex. In THF,
CO2 reduction selectively leads to CO formation, whereas
experiments in acetonitrile exclusively afford formate, HCOO−.
In order to explain the experimental findings, theoretical
calculations of the reaction pathways were performed and further demonstrate the importance of the applied solvent for a
selective reduction of CO2.
−
diketiminate)Fe(μ-N)]2 leads to formation of CO3 and
CO.6 A selective catalytic hydrogenation of CO2 to formic
acid was achieved utilizing trans-[(tBu-PNP)Fe(H)2(CO)] (3)
at 6.66 bar of H2 and 3.33 bar of CO2 pressure in H2O/THF
mixtures with turnover numbers of up to 788.7 When
modifying the PNP ligand using a 2,6-diaminopyridine scaffold
instead of 2,6-dimethylpyridine (4, Figure 1), a significant
increase of the turnover numbers of up to 10275 was reported,
although at much higher temperatures and pressures, 80 °C
and 80 bar of H2/CO2 (1:1).8 An efficient catalytic
hydrogenation of CO2 was also reported utilizing [Si(o-
INTRODUCTION
■
Carbon dioxide is one of the major greenhouse gases, and its
activation is an important area owing to its scientific and social
relevance. However, only a handful of catalytic systems have
been reported that allow for an efficient conversion of this
greenhouse gas. The majority of these systems are based on
precious metals such as Ru and Ir, rendering them unsuitable
for large-scale sustainable applications due to their high price
and low natural abundance. Consequently, more abundant
alternatives based on Fe, Co, and Ni have recently attracted
more attention with Fe based systems being most prominent.
While heterogeneous systems based on Fe often result in a
broad variety of products, e.g., CO, formic acid, and methanol,
homogeneous catalysts usually lead to a more narrow product
distribution and can thus serve to provide an in-depth
understanding of the underlying mechanistic details.1,2 One
of the most potent classes of Fe-based complexes are Fe-
hydrides, which allow for an efficient hydrogenation of CO2 to
formate (Scheme 1). A prominent example is Fe(dmpe)2(H)2
(1, dmpe = 1,2-bis(dimethylphosphino)ethane) that was
shown to effectively reduce CO2 to formate.3,4 Likewise, low-
coordinated diiron [(β-diketiminate)Fe(μ-H)]2 (2) and its
derivatives react stoichiometrically with CO2 in pentane at low
temperatures to afford η2-formato bridging diiron complexes.5
On the other hand, the reaction of CO2 with [(β-
C6H4PR2)3)Fe(H2)(H)] (5, R = Pr, Ph) or [(PhBPiPr3)Fe-
i
(H)3(PMe3)] (6) at elevated temperatures and pressures of
CO2 and H2. For both complexes, formation of (Et3NH)-
(OCHO) and MeOCHO was observed in methanol at 100 °C
and a total pressure of 58 bar of H2/CO2 (1:1) in the presence
of trimethylamine.9 While TriphosC/Fe(BF4)2 mixtures
(TriphosC = CH3C(CH2PPh2)3) were recently reported to
be unreactive toward the reduction of CO2,11 [(TriphosC)Ru-
(acetate)] (7) is an active catalyst for its conversion and
affords methanol with turnover numbers of up to 442 at a
CO2/H2 pressure of 80 bar and 140 °C.8 Notably, the solvent
Received: October 1, 2018
© XXXX American Chemical Society
A
DOI: 10.1021/acs.organomet.8b00711
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Scheme 1. Selected Literature Reported Metal-Hydride
3−10
Complexes for the Conversion of CO2
intermediate [(TriphosC)Ru(η2-O2CH)(solvent)] was sug-
gested as a key intermediate in this transformation.
Furthermore, [(TriphosCCu)2(H)] (8) was reported to allow
for reduction of CO2 to afford formic acid.10
We recently showed that (Triphos)Fe complexes exhibit an
unusual solvent-dependent κ2- vs κ3-binding flexibility of the
Triphos ligand leading to an altered reactivity of the
structurally equivalent complexes [(TriphosC)FeCl2] and
[(TriphosSi)FeCl2].12 The previously observed strong sol-
vent-dependent reactivity of such iron complexes prompted us
to investigate the reaction of CO2 and (TriphosC/Si)-derived
Fe-hydride complexes in different solvents. Furthermore,
[(TriphosC)Fe(H)(BH4)] was previously shown to possess a
temperature-dependent flexible H−/H2 bonding of the BH4
moiety to iron.13,14 Inspired by these observations, we focused
on the reaction of [(TriphosC)Fe(H)(BH4)] as a well-defined
Triphos−Fe hydride with CO2. We herein report on its
solvent-controlled reaction with CO2, leading exclusively to the
formation of either CO or formate. A detailed spectroscopic
investigation together with extensive computational studies
rationalizes the experimental observations on a molecular level.
Figure 1. (a) UV/vis spectra of complex 11 in THF and acetonitrile
as well as after reaction with CO2 under various reaction conditions.
(b) ATR-IR spectra of complexes 13 (black), 14 (red), and 15 (blue).
̈
27 FT-IR spectrometer. Zero-field Mossbauer spectra were recorded
at 80 K by using a SeeCo constant acceleration spectrometer
equipped with a temperature controller maintaining temperatures
within 0.1 K and a 57Co radiation source in a Rh matrix. Isomer
shifts are referred to α-Fe at room temperature. Data were fit with a
sum of Lorentzian quadrupole doublets by using a least-squares
routine with the WMOSS program. Head space analysis of the
gaseous components was performed utilizing a 7820A GC system
from Agilent Technologies equipped with a methanizer from Joint
Analytical Systems, a flame ionization detector (FID), and a thermal
conductivity detector (TCD). The system is equipped with a
HaysepQ and molesieve 5 Å column, and argon is used as the carrier
gas. Quantification of formic acid was performed by HPLC on a C18
pyramid column applying 0.5 M H3PO4 as the mobile phase.
EXPERIMENTAL SECTION
■
General. All reactions were performed under a dry N2 or Ar
atmosphere, either using standard Schlenk techniques or by working
in a glovebox. All solvents applied were dried according to standard
methods. The tripodal phosphine ligands TriphosC and TriphosSi 15 as
well as their iron complexes [(TriphosC)FeCl2] (9) and [(TriphosSi)-
FeCl2] (10)12 were synthesized according to literature procedures. All
other compounds were obtained by commercial vendors and used
without further purification. UV−vis measurements were performed
with a Varian Cary 100 or a Jasco V-670 utilizing an inert cuvette with
a septum seal. 1H and 13C NMR spectra were recorded with a Bruker
DPX-250 NMR spectrometer at room temperature under inert
conditions. ESI mass spectra were obtained with a Bruker Daltronics
Esquire6000 instrument. GC-MS measurements were performed with
a Shimadzu QP-2010. IR spectra were measured with a Bruker Tensor
Computational Details. All calculations were performed using
the ORCA program package.16,17 In the course of our computational
study, multiple sets of calculations on different levels of theory were
conducted. The first set comprises relaxed surface scans and geometry
optimizations that employ the BP86 functional18,19 together with the
def2-SVP double-ζ basis set.20 These calculations helped to explore
possible reaction pathways without investing immense computational
resources. Final geometries for stable structures and transition states
were computed using the B3LYP functional21,22 and the def2-TZVP(-
f) basis set.20 Integral generation was accelerated with the resolution
of identity and chain of sphere (RIJCOSX) approximations in
conjunction with the appropriate auxiliary basis set.23−28 Dispersion
effects were accounted for using the semiempirical van der Waals
B
DOI: 10.1021/acs.organomet.8b00711
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
−1
̈
corrections with the Becke−Johnson damping (D3BJ) as imple-
mented in ORCA.29,30 While the same computational setup was
utilized for the calculation of electronic absorption spectra, we have
supplemented it with the zeroth order regular approximation (ZORA)
1435, 1185, 1092. Mossbauer: δ = 0.11 ( 0.02) mm s , ΔEQ = 1.96
( 0.02) mm s−1. ESI-MS: calcd for [C41H44BFeP3]+, 696.22; found,
696.10. Anal. Calcd for C41H44BFeP3: C, 70.72; H, 6.37. Found: C,
70.5; H, 6.2.
[(TriphosC)FeD(BD4)] (11-D). The deuterated complex 11-D was
obtained following the same synthetic route as that described for
complex 11 utilizing NaBD4. Purple crystals of 11-D were obtained by
slow diffusion of pentane into a solution of 11-D dissolved in THF.
[(TriphosSi)Fe(H)(BH4)] (12). TriphosSiFeCl2 (1 g, 1.3 mmol) was
dissolved in dry THF (40 mL) followed by the addition of sodium
borohydride (493 mg, 13.0 mmol). The obtained dark red solution
was then stirred for 2 d at room temperature. After filtration of the
reaction mixture and concentrating the solution to 3 mL, complex 12
was precipitated by addition of Et2O (60 mL). Filtration of the
formed purple solid, washing with Et2O, and drying in a vacuum
yielded complex 12 in 19% yield (179 mg, 0.25 mmol). Crystals of 12
were obtained by slow diffusion of pentane into a THF solution of
and the scalar-relativistic recontractions of the same basis sets during
31,32
̈
the calculation of Mossbauer isomer shifts.
In the case of
electronic absorption spectra, the underlying electronic excitation
energies and transition moments were evaluated within the framework
of time-dependent linear response density functional theory (TD-
DFT). Eventually, electronic energies were refined at high precision
using the DLPNO-CCSD(T) method.33−36 The corresponding
coupled cluster theory with single, double, and perturbative triple
excitations CCSD(T) is considered as the “gold standard” owing to its
high accuracy and reliability. Its domain based local pair natural
orbital coupled cluster version DLPNO-CCSD(T) typically repro-
duces canonical CCSD(T) energies with chemical accuracy but
comes at immensely reduced computational cost. The presented
numbers were obtained from a two-point extrapolation to the
complete basis set limit from the energies computed with the CC-
PVTZ and CC-PVQZ basis sets.37−42
1
compound 12. H NMR (CDCl3, 250 MHz, ppm): δ 7.45−6.99 (m,
30H, HAr), 1.60 (d, 6H, CH2), 0.56 (s, 3H, CH3), −12.97 (1H, s,
hydride); IR (ATR, cm−1): 3422, 3052, 2926, 2390, 2293, 2226,
All geometries were fully optimized without symmetry constraints.
Harmonic vibrational frequencies were computed to verify the nature
of stationary points. Each of the intermediates reported is a local
minimum on the potential energy surface, and therefore, the
corresponding Hessian matrix has exclusively positive eigenvalues,
whereas the Hessian matrices of transition states exhibit a single
negative eigenvalue. Contributions from translational, rotational, and
vibrational degrees of freedom to the molecular enthalpy and entropy
were accounted for within the “particle in a box”, “rigid rotor”, and
“harmonic” approximations.43,44
1889, 1636, 1481, 1434, 1383, 1186, 1127, 1085, 1000, 820, 768, 744,
−1
̈
696. Mossbauer: δ = 0.14 ( 0.02) mm s , ΔEQ = 1.90 ( 0.02) mm
s−1. Anal. Calcd for C40SiH44BFeP3: C, 67.43; H, 6.23. Found: C,
67.0; H, 6.1.
[(TriphosC)Fe(CO)2] (13). Compound 11 (350 mg, 0.503 mmol)
was dissolved in dry THF (28 mL) in a 100 mL stainless-steel
autoclave and pressurized with CO2 to give an internal pressure of 7
bar at 70 °C. Subsequently, the mixture was stirred under such
conditions for 2 d. After opening the pressurized bottle
(ATTENTION: Instantaneous gas release should be avoided, and
the autoclave should solely be opened under well-ventilated
conditions), the obtained dark green solution was evaporated to
dryness under the exclusion of air. The solid residue was then
suspended in dry THF (5 mL), filtrated, and washed with hexane to
yield 13 (290 mg, 0.387 mmol, 78%) as a dark green solid. Green
crystals were obtained by slow diffusion of pentane into a THF
A key factor in the observed reactivity is the influence of the
solvent. In our calculations, it is modeled by a combination of implicit
and explicit solvation: While unspecific interactions with the reactive
site are evaluated with a polarizable continuum model (C-PCM,45)
specific interactions such as chemical bonding to Fe are taken into
account by explicitly including solvent molecules in the model system
(see discussion below). For intermediates immediately before binding
or elimination of a solvent or CO2 molecule also unbound molecules
had to be included in order to obtain a consistent picture of the
reaction pathway. This inevitably leads to a considerable under-
estimation of the entropy contributions to the free energy, as the
individual translational entropy of the ensemble of individual
molecules is (wrongly) treated as translation of a single entity.
Therefore, we compared our results for reaction steps that feature
addition or elimination of a molecule with results from separate
calculations for each of the reactants. In this way, an upper bound to
the error is obtained (Table S17). To this end, it should be noted that
for technical reasons the C-PCM model was only applied during final
energy evaluations but not during geometry optimizations. However,
a series of test calculations revealed that the neglect of implicit
solvation does not alter geometries or energies significantly (Table
S18). Moreover, the data presented in Tables S15 and S16
demonstrate that the computed structures closely agree with the
available crystal structures, thus validating the chosen setup. A
pressure correction value of 1.34 kcal mol−1 per molecule of CO2 is
added while plotting the free energy profile for the reactions with
pressurized CO2 (Supporting Information T1).
solution of 13. IR (ATR): ν
̃
(cm−1) = 3053, 2959, 2925, 2386, 1907,
̈
1796, 1742, 1580, 1513, 1425, 1091, 1059, 879. Mossbauer: δ = 0.13
( 0.02) mm s−1, ΔEQ = 0.26 ( 0.02) mm s−1. Anal. Calcd for
C43H39FeO2P3: C, 70.12; H, 5.34. Found: C, 69.9; H, 5.4.
Samples of compound 13 can also be prepared from TriphosC and
[(COT)Fe(CO)3] following the route described by Lindner et al.
earlier.46
Generation of [(TriphosC)Fe(H)(OOCH)] (14). Compound 11
(150 mg, 0.216 mmol) was dissolved in dry acetonitrile (20 mL).
CO2 was then purged through the solution, resulting in an instant
color change from purple to bright orange. After additional stirring for
30 min at room temperature under a CO2 atmosphere, the obtained
mixture was filtrated and the solvent was removed in a vacuum to
1
yield 80 mg of [(TriphosC)Fe(H)(OOCH)]. H NMR (250 MHz,
CD3CN) = 8.65 (s, OOCH), 7.30−6.96 (m, HAr), 2.36 (d, CH2),
1.64 (s, CH3), −12.07 (s, hydride). IR (ATR): ν
̃
(cm−1) = 3055,
2384, 1710, 1580, 1479, 1431, 1180, 1095, 1060, 995, 836, 735, 691.
−1
Mossbauer: δ = 0.09 ( 0.02) mm s , ΔEQ = 1.33 ( 0.02) mm s−1.
̈
Despite repeated attempts, it was not possible to obtain satisfactory
elemental analyses.
Generation of [(TriphosC)Fe(OOCH)2] (15). Compound 11 (1
g, 1.437 mol) was dissolved in dry acetonitrile (35 mL) in a 100 mL
stainless-steel autoclave and pressurized with CO2 to show an internal
pressure of 7 bar at 75 °C. The mixture was then stirred under such
conditions for 2 d. After opening the pressurized bottle
(ATTENTION: Instantaneous gas release should be avoided, and the
autoclave solely being opened under well-ventilated conditions), the
obtained blue solution was evaporated to dryness. The solid residue
was then suspended in dry acetonitrile (5 mL), filtrated, and washed
with hexane to yield 780 mg of [(TriphosC)Fe(OOCH)2] as a blue
[(TriphosC)Fe(H)(BH4)] (11). TriphosCFeCl2 (1 g, 1.33 mmol)
was dissolved in dry THF (40 mL), and sodium borohydride (508
mg, 13.3 mmol) was added in one portion. The resulting dark red
solution was stirred at room temperature for 2 d and subsequently
filtered. The filtrate was then concentrated to 3 mL by slow
evaporation in a vacuum, and addition of hexane (60 mL) afforded
compound 11 as purple solid that was filtered off, washed with
hexane, and dried in a vacuum (833 mg, 1.2 mmol, 90%). Purple
crystals of 11 were obtained by slow diffusion of pentane into a
1
solution of 11 dissolved in THF. H NMR (250 MHz, CDCl3): δ
solid material. IR (ATR): ν
1479, 141, 1088, 832, 731, 691. Mossbauer: δ = 0.13 ( 0.02) mm s ,
̃
(cm−1) = 3050, 2988, 2379, 1696, 1588,
7.59−7.19 (m, 30H, HAr), 5.89 (br s, 2H, terminal BH4), 2.52 (s, 6H,
−1
̈
CH2), 1.90 (s, 3H, CH3), −4.82 and −14.18 (s, 3H, hydride); IR
(ATR): ν
̃
(cm−1) = 3053, 2922, 2390, 2294, 2226, 1881, 1629, 1481,
ΔEQ = 0.26 ( 0.02) mm s−1. ESI-MS: calcd for [C43H41FeO4P3 +
C
DOI: 10.1021/acs.organomet.8b00711
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Na]+, 793.15; found, 793.36. Anal. Calcd for C43SiH41FeO4P3: C,
67.03; H, 5.36. Found: C, 66.6; H, 5.6.
[(TriphosC)Fe(μ-H)3Fe(μ-H)3Fe(TriphosC)] and [Tri-
phosC(BH3)3] (Figure S4), were only observed when complex
11 was dissolved in polar solvents (e.g., acetonitrile) and
stored over an extended period of time (>1 week). In dry and
oxygen free nonpolar solvents (e.g., THF, hexanes), complex
11 did not show any spectroscopic changes (Figure S5a),
although rapid decomposition is observed (Figure S5b) upon
exposure to oxygen. The difference in reactivity of 11 and 12
can be explained by the altered electronics caused by the C/Si
exchange and stems from a geometrical control of the ligand
field splitting.12,50 This difference is best displayed by their
RESULTS AND DISCUSSION
■
[(TriphosC/Si)Fe(H)(BH4)]. The synthesis of [(TriphosC)-
Fe(H)(BH4)] was previously reported via the reaction of
TriphosC with [Fe(H2O)6](BF4)2 and NaBH4.13,47 In contrast,
utilization of FeCl2, FeBr2, and [Fe(H2O)6](ClO4)2 salts was
reported to afford [(Triphos)Fe(μ-H)3Fe(Triphos)]+.48,49
Due to the dependence on the applied reaction conditions,
we first focused on a straightforward reproducible synthetic
method to generate [(TriphosC)Fe(H)(BH4)] (11) and its Si
counterpart, [(TriphosSi)Fe(H)(BH4)] (12), on the gram
scale (Scheme 2).
̈
Mossbauer parameters. While complex 11 reveals an isomeric
shift of δ = 0.11 mm s−1 and a quadrupole splitting of ΔEQ =
1.96 mm s−1, compound 12 shows an isomeric shift of δ = 0.14
mm s−1 and a quadrupole splitting of ΔEQ = 1.90 mm s−1
(Figure S6). Here, the smaller isomer shift of 12 points at a
slightly increased s-electron density at the Fe center and is in
line with previous reports on TriphosC/TriphosSi derived iron
complexes.12 The change in the electron distributions caused
by the different ligand platforms is further supported by the
shifts of the Fe−H 1H NMR signals observed in CDCl3. Here,
the Fe-hydride resonance for 11 is observed at −14.18 ppm
concomitant with an additional borohydride signal at −4.82
ppm (Figure S7a). Under identical conditions, the correspond-
ing spectrum of 12 solely shows a Fe−H signal that is low field
shifted to −12.97 ppm and further suggests an increased s-
electron density at the iron center (Figure S9).
Scheme 2. Synthesis and Reactivity of Triphos-Derived Iron
Hydride Complexes
Hydride Transfer. Since [(TriphosSi)Fe(H)(BH4)] (12)
rapidly rearranges to [(TriphosSi)Fe(μ-H)3Fe(μ-H)3Fe-
(TriphosSi)], which lacks a reliable synthetic purification
method, it evades being a potential platform for hydrogenation
reactions. Therefore, we focused on the reactivity of complex
11 and first tested its potential to facilitate hydride transfer to
simple organic molecules. Here, stoichiometric hydrogenation
of nitrobenzene as well as benzophenone to aniline and
diphenylmethanol, respectively, was achieved in quantitative
yields (Supporting Information: Stoichiometric hydrogena-
tion). A catalytic conversion, however, was not achieved, and
addition of H2 (1 atm), Et3SiH, BH3, NaBH4, or N2H4 did not
allow for more than one turnover.
CO2 Reduction in THF. Inspired by the reactivity of 11
toward keto and nitro groups in organic molecules, we next
tested the complexes’ potential for the hydrogenation of CO2.
While 11 did not react with CO2 in THF under atmospheric
conditions, conversion of CO2 was readily observed at 7 bar
and 70 °C visible by a distinct color change from purple to
dark green. The UV/vis absorption bands change accordingly
from 515 nm in complex 11 (Figure 1a) to afford an almost
featureless spectrum with weak absorption bands at 612 and
805 nm upon reaction with CO2. Crystallization of the product
mixture from THF/pentane afforded green crystals which
revealed a distinct IR band at 1907 cm−1 (Figure 1b). This
characteristic CO stretching frequency suggests a conversion of
CO2 to afford metal-bound CO. Crystallographic analysis
revealed the molecular structure of complex 13 (Figure 2,
Table 1). Here, the TriphosC ligand is bound to the central
iron in a κ3-mode with equal P−Fe distances of 2.20 Å. Two
additional CO ligands complete the coordination sphere. The
Fe−CO distances were found to be 1.75 Å. According to the
molecular structure, the Fe-center can be referred to as a Fe(0)
We found that the reaction of [(TriphosC)FeCl2] (9) and 10
equiv of NaBH4 exclusively leads to [(TriphosC)Fe(H)(BH4)]
(11) in 90% yield. The molecular structure (Figure S1)
confirms the expected composition showing a κ3-coordinated
Triphos ligand with iron as well as an additional hydride and a
−
bidentate BH4 ligand. Notably, fast interchange between the
−
Fe−H and the bridging hydrogen atoms of the BH4 moiety
was previously reported on the basis of NMR spectroscopy,
suggesting an overall similar reactivity of the terminal and
bridging hydrides in 11. While the synthesis of the isostructural
Si-derived compound 12 (Scheme 2, Figure S2) follows an
identical reaction protocol, the desired complex is only
obtained in 19% yield, with the major products being
[(TriphosSi)Fe(μ-H)3Fe(μ-H)3Fe(TriphosSi)] and [Tri-
phosSi(BH3)3] (Figure S3) along with other hitherto
unidentified Fe species. The respective C-analogues,
̈
species. This finding is also in line with the Mossbauer
spectrum of complex 13 (Figure S15), revealing an isomer shift
of 0.13 mm s−1 with a quadrupole splitting of 0.26 mm s−1.
D
DOI: 10.1021/acs.organomet.8b00711
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Upon change of the solvent from THF to acetonitrile, the
main absorption band of complex 11 shifts from 515 to 525
nm (Figure 1a). While this shift might be due to implicit
solvent effects (e.g., change in electric field, etc.), we believe
that it originates from coordination of an additional solvent
molecule to the Fe center (see the computational section
1
below). H NMR spectroscopy further supports an altered
electronic distribution. Here, the hydride signal at −14.18 ppm
in nonpolar solvents (CDCl3 and THF-d8, Figure S7a,b) shifts
to −10.06 ppm in CD3CN (Figure S7c). Notably, a significant
1
increased broadening of the borohydride H NMR signal was
observed in acetonitrile. Notably the line width significantly
varies as a function of temperature (Figure S8) and suggests a
change of the binding of the H atoms as well as a likely
coordination of the solvent. Similar line broadenings were
previously reported for (PNMeP)Fe(H)BH4 (PNP = HN-
{CH2CH2(PR2)) and suggest an overall alteration of the
Figure 2. Molecular structure of complex 13 with thermal ellipsoids
drawn at the 50% probability level (hydrogen atoms and ellipsoids of
the phenyl rings are omitted for clarity). Gray, carbon; purple,
phosphorus; blue, iron; red, oxygen.
Table 1. Selected Bond Lengths and Angles of Complex 13
58
binding mode of this ligand from κ2 to a κ1 bound BH4 .
−
bond length (Å)
While the reaction of 11 and CO2 under ambient pressure
and room temperature did notably not proceed in THF
solutions, the analogous reaction in acetonitrile instantly led to
a color change of the reaction mixture from purple to orange
along with a shift of the band from 525 to 464 nm (Figures 1a
and S15). Whereas a faster reaction with CO2 might be
expected in acetonitrile due to the significantly increased
solubility of CO2 in acetonitrile versus THF,59 the CO2
reduction product was likewise altered visibly by a new IR
band at 1710 cm−1 in the reaction mixture as well as isolated
complex 14 (Figure 1b). This band is indicative for the
formation of formate in acetonitrile under ambient conditions.
Notably, GC and HPLC analysis (Figure S18) confirmed the
sole formation of formate under such conditions and thus
shows a significantly different product formation as observed in
THF. While we were not able to obtain single crystals suitable
Fe1−P1
Fe1−P2
Fe1−P3
Fe1−C1
Fe1−C2
2.2015(6)
2.2007(6)
2.1940(6)
1.752(2)
1.756(2)
bond angle (deg)
P1−Fe1−P2
P1−Fe1−P3
P2−Fe1−P3
P1−Fe1−C1
P1−Fe1−C2
P2−Fe1−C1
P2−Fe1−C2
P3−Fe1−C1
P3−Fe1−C2
C1−Fe1−C2
86.34(2)
92.78(2)
94.19(2)
89.69(8)
156.50(9)
110.62(9)
152.63(8)
86.75(8)
112.57(8)
86.22(11)
̈
for X-ray structure determination, Mossbauer spectroscopy
gave additional evidence for the formation of complex 14. The
isomer shift of 0.09 mm s−1 together with a quadrupole
splitting of 1.33 mm s−1 is in accordance with the formation of
Notably, the synthesis of complex 13 was previously reported
via reaction of TriphosC and (COT)Fe(CO)3, and the
spectroscopic data obtained via this method are in good
accordance with those of complex 13 generated via CO2
reduction.46 In addition, GC as well as HPLC analyses were
performed to determine any potential CO2 reduction products
in the headspace or the reaction solution. Here, CO was solely
detected in the headspace of as well as within the reaction
mixture and thus highlights the specificity of the CO2
reduction toward CO (Figure S15).
1
a low spin Fe(II) complex (Figure S19). In addition, the H
NMR spectrum of complex 14 in acetonitrile-d3 revealed a
signal at −12.07 ppm, indicating that the Fe center is
potentially coordinated by an additional hydride ligand as
well as a formate ligand (Figure S20).
H/D Exchange. The formation of formate from CO2 was
further supported by utilizing the deuterated complex,
[(TriphosC)Fe(D)(BD4)] (11-D), during the reduction
experiments. While the results were comparable to the IR
pattern of 14, the band at 1710 cm−1 (14) was shifted to 1684
cm−1 (Figure S21), which is in accordance with the expected
H/D exchange in the formate ligand. The high stretching
frequency of the CO group as well as the appearance of the
CHformate resonance at 8.65 ppm furthermore supports the
formation of an iron bound formate molecule (Figure S20).58
Thus, the formate formed is clearly a result of CO2 reduction
facilitated by complex 11. Notably, while the neat reaction of
CO2 and NaBH4 in acetonitrile was previously reported to
afford [HB(OCHO)3]− and [H2B(OCHO)2]−,60,61 neither
11B NMR nor IR spectroscopy revealed the typical signals for
such species. Furthermore, to investigate the possibility of a
CO2 Reduction in Acetonitrile. While stable in THF,
complex 11 slowly converts into [(Triphos)Fe(μ-H)3Fe(μ-
H)3Fe(Triphos)] in acetonitrile over the course of 1 week.
The different behavior of 11 in acetonitrile pointed our
attention toward a potentially different reduction pathway of
CO2 in another solvent. Moreover, the solvent polarity was
previously reported to influence the hydricity of transition
metal hydrides51−57 and an increase of the hydride donor
strength of a metal complex corresponding with an increasing
solvent polarity was shown. Along this line, a solvent-
dependent CO2 hydrogenation was reported for H2Co-
(dmpe)2.57 Herein, depending on the used solvent severe,
−
alteration of the reaction pathway to afford HCO2 was
−
direct CO2 reduction by dissociated BH4 followed by rapid
observed, while the product distribution was still the same. On
the basis of such earlier reports, we anticipated that the solvent
change will play a major role for the hydrogenation of CO2.
formate interception by the Fe complex, the reaction of 11
with formate was attempted but neither agrees with the
E
DOI: 10.1021/acs.organomet.8b00711
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Scheme 3. Proposed Reaction Mechanisms for the Reaction of 11 and CO2 in Acetonitrile at Room Temperature and 1 atm of
a
CO2 (top) and THF at 70 °C and 7 bar of CO2 (bottom)
a
All spectroscopically characterized species (see text) are boxed. Note that, in the case of species B, only two isomers are observed experimentally
(see Scheme 4 and the corresponding text).
spectrum of 15, 14, nor compound B (Scheme 3) and thus
further suggests a purely metal centered reduction process.
The Importance of Pressure. To exclude the different
pressure conditions as the reason for the varied product
formation, the reaction of 11 with CO2 in acetonitrile was
repeated at 7 bar/70 °C. Although a different product
formation can be observed, no CO could be detected. Instead
of the IR pattern of 14, the spectrum of the reaction mixture of
11 with CO2 in acetonitrile at 7 bar/70 °C reveals two bands
at 1696 and 1589 cm−1 that can be reasoned by the formation
of compound 15 (Figure 1b). Furthermore, when a solution of
sodium borohydride in acetonitrile was purged with CO2
under otherwise identical experimental conditions as those
used for the formation of 14, the IR spectrum of the formed
colorless solid clearly differs from the one observed for
complexes 14 and 15 (Figure S22). This finding furthermore
shows that an iron centered reaction mechanism for the
formate formation has to be considered and not the reduction
molecule. Importantly, the bulky phenyl groups form a
shielded cavity in which all reactions take place. In contrast
to previous findings for structurally related compounds in
different solvents,12 a change of the binding mode of the
Triphos ligand was found to be unfavorable (Figure S40). The
abundance of hydrides in 11 in principle opens up possibilities
for the formation of H2 and for several low energy pathways
during the course of the reaction. However, the reactivity of
the Fe-bound hydride is limited by its poorly accessible
position within the cavity, while the two peripheral hydrides
are bound to the electron-deficient boron. Accordingly, a direct
insertion in the Fe−H or B−H bonds of 11 by CO2 is
associated with prohibitively large reaction barriers (Figure
S39). Instead, the reaction in acetonitrile is initiated by binding
of a solvent molecule to a Fe(II) center, resulting in A.
Experimental support for this finding is provided by the good
agreement between the UV/vis spectrum of 11 after
dissolution in acetonitrile and the calculated spectrum of A
(Figures S23 and S25) as well as a broadening of the 1H NMR
(vide supra). Concomitant to the binding of acetonitrile to Fe,
the borohydride changes its coordination mode from κ2 to κ1
which increases the number of peripheral hydrides from two to
three and renders each of them more nucleophilic. This
effectively reduces the barrier for an electrophilic attack by
CO2 to ΔG⧧(TS-A-B) = 11.6 kcal mol−1, and accordingly, the
formate containing B is readily formed (Scheme 3). An
analogous reaction in THF is prevented by steric factors.
Owing to its relatively large size, a THF molecule cannot enter
the cavity of 11 and consequently no activation of the
peripheral hydrides occurs. If, however, the temperature and
pressure are increased, CO2 acts as a nucleophile and attacks
the boron before binding to the Fe via its carbon (Scheme 3)
which is a prerequisite for CO formation.2,62−64
−
of CO2 by free BH4 due to decomposition of 11.
Computational Studies. In order to rationalize the
unique solvent-controlled CO2 reduction, the different
reactions were investigated by means of quantum chemical
calculations. In the course of this investigation, a large number
of possible reaction pathways were simulated. For the sake of
clarity, we herein constrain the discussion to the energetically
most favorable pathways as well as some particularly relevant
alternatives and refer the reader to the Supporting Information
(Supporting Information: T3) for additional information. The
most favorable mechanisms are shown in Scheme 3.
Before giving a detailed description of the full reaction
mechanisms, we start with a brief characterization of the
starting material 11 and focus the discussion on the initial step
of the reaction that is responsible for the different product
selectivity in different solvents. According to our results, 11 has
a closed shell singlet electronic ground state with a central
Fe(II) ion. Close agreement between experimentally obtained
data and calculated values for key structural parameters (Table
The first step of the reaction in acetonitrile corresponds to
one solvent molecule binding to the Fe(II) center in complex
11, resulting in A. As depicted in Figure 3, this step is
associated with a barrier of ΔG⧧(TS-11-A) = 8.4 kcal mol−1
and by ΔG(A/11) = 6.4 kcal mol−1 endothermic. Never-
theless, the abundance of acetonitrile together with the
following irreversible and exothermic addition of a CO2
molecule to A act as a driving force for this step. The
reduction of CO2 by one of the peripheral hydrides in A results
in the formation of a metastable intermediate with a κ2-bound
borohydride and an unbound formate anion.
̈
S16) and Mossbauer isomer shifts (Table S15) as well as good
agreement between the experimental and calculated UV/vis
spectra (Figures S23 and S24) indicate that our electronic
structure calculations adequately describe the investigated
system. The facial coordination of the Triphos ligand in 11
imposes a favorable geometrical arrangement such that all
observed reactions occur at the borohydride face of the
F
DOI: 10.1021/acs.organomet.8b00711
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
with a κ1 bound formate while according to our calculations
changing the binding mode to κ2 would lead to the more stable
14 (ΔG(14/B4) = −7.5 kcal mol−1). In contrast to the
̈
Mossbauer and IR spectra, the UV/vis spectrum was recorded
after the produced solid was again dissolved in acetonitrile.
Therefore, it is not unexpected that the calculated spectrum of
B3 fits the experimental spectrum best (Figures S26 and S29).
Thus, we believe that, although B2 is more stable than B, the
latter is the predominant species in an acetonitrile solution.
Moreover, for the same reason as that discussed above for 11
and A, B is more reactive toward a nucleophile attack than B2.
Upon applying a CO2 pressure of 7 bar and increasing the
temperature to 70 °C, B reacts with a second molecule of CO2.
The corresponding reaction profile is presented in Figure 4. It
Figure 3. Calculated free energy profile for the reaction of 11 with
CO2 in acetonitrile at ambient temperature and pressure (r.t./1 atm of
CO2). All values are given in kcal mol−1 and were obtained at the
DLPNO-CCSD(T)/CBS level of theory (see the Computational
Details section).
Due to the prior activation of the peripheral hydrides, this
step has a relatively low barrier of ΔG⧧(TS-A-B) = 11.6 kcal
mol−1 as compared to the addition of CO2 to unactivated 11.
The metastable intermediate with an unbound formate further
undergoes a geometrical reorganization where the formate
binds to the Fe(II) center while breaking one Fe−H−B bond
in a concerted fashion, which yields the stable product B
(ΔG(B/11) = −6.5 kcal mol−1). The predicted low barriers for
the formation of B are in agreement with the experimental
findings of a facile reaction. Under the given reaction
conditions, B may eliminate the Fe-bound acetonitrile to
give B2 (Scheme 4), which is more stable by ΔG(B2/B) =
Figure 4. Calculated free energy profile for the reaction of 11 with
CO2 in acetonitrile at 70 °C and a pressure of 7 bar. All values are
given in kcal mol−1 and were obtained at the DLPNO-CCSD(T)/
CBS level of theory (see the Computational Details section).
Scheme 4. Structures of Possible Isomers of Compound B
Together with Their Calculated Relative Stability in kcal
mol−1
must be noted that, although both the final point in Figure 3
and the first point in Figure 4 correspond to compound B,
their absolute energies differ by 10.0 kcal mol−1, as they
correspond to a calculation including the second CO2
molecule (ΔG(B/11) = 3.5 kcal mol−1, Figure 4) and two
separate calculations of complex and substrate (ΔG(B′/11) =
−6.5 kcal mol−1, Figure 3).
The difference between the two free energies largely
originates from differences in the translational entropy (see
the Computational Details section). Starting from the former,
our calculations indicate that the second CO2 approaches the
singly bound borohydride and reacts in an SN2 fashion to
increase the electropositive character of the carbon. Sub-
sequently, this carbon inserts into the just formed Fe−H bond
to give C (Figure 4). Despite intense efforts, the optimization
of the transition state TS-B-C did not quite meet the
convergence criteria. Nevertheless, the energy during opti-
mization steps only changes marginally (ΔE ≈ 0.1 kcal mol−1)
and the highest imaginary harmonic frequency for the partially
converged configuration is associated with a symmetric mode
of the SN2 fashioned reaction with boron being the central
−8.9 kcal mol−1. Furthermore, we assume that evaporation of
the solvent prior to the spectroscopic characterization of the
reaction product effectively leads to the elimination of both,
acetonitrile and BH3 (as an acetonitrile adduct), since the
experimental isomer shift (Figure S19) and CO stretching
frequency (1710 cm−1) only match the calculated values for
intermediate B4 (Scheme 4, Tables S7−S9, and Table S14).
Interestingly, B4 features a 5-fold coordinated Fe(II) center
G
DOI: 10.1021/acs.organomet.8b00711
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Figure 5. Calculated free energy profile for the reaction of 11 with CO2 in THF at 70 °C and 7 bar of CO2. All values are given in kcal mol−1.
atom. Therefore, we strongly believe that the obtained
structure resembles the true transition state and the calculated
barrier for the same is an upper bound to the true reaction
barrier. Even though the partial formation of the C−H bond
stabilizes the transition state, the associated activation barrier is
comparably large (ΔG⧧(TS-B-C) = 27.1 kcal mol−1),
consequently making TS-B-C a kinetically inaccessible
conformation at ambient temperature and pressure. In an
alternative pathway that would feature a direct insertion of
CO2 in the B−H bond, no stable product could be located
(Figure S38). We attribute this observation to the reduced
nucleophilic character of the peripheral hydrides in B as
compared to the extremely electron-rich compound A.
Importantly, only after elimination of acetonitrile and BH3
(as an acetonitrile adduct) from C to obtain 15, the total
reaction free energy becomes negative (ΔG(15/B) = −8.9 kcal
mol−1). Good agreement between the experimental and
calculated CO stretching frequencies (exptl 1589/1696 cm−1
vs calcd 1605/1707 cm−1, Table S11) as well as the UV/vis
spectrum (Figures S31 and S32) support the identity of 15 as
the final reaction product.
reaction mechanism. According to our results, the first step
of the reaction involves the insertion of CO2 in one of the two
electron-deficient three-center-two-electron B−H−Fe bonds
to give intermediate E that features a stable six-membered ring
(Scheme 3). As expected, this endothermic (ΔG(E/11) = 10.6
kcal mol−1) step has a high barrier of ΔG⧧(TS-11-E) = 37.3
kcal mol−1 and is hence only proceeded through high
temperatures and pressures (Figure 5). After its formation, E
isomerizes to a Fe(0)−H2 species with a short H−H bond of
0.83 Å. This change in oxidation state at the Fe is critical for
the observed overall reaction, as it allows the Fe to act as a
nucleophile and form a Fe−C bond (F). Despite the
considerable free energy gain during this step (ΔG(F/E) =
−11.9 kcal mol−1), no transition state could be localized, since
the corresponding part of the potential energy surface
calculated with the setup used for geometry optimizations is
extremely flat (ΔGDFT(F/E) = −1.1 kcal mol−1). We are thus
convinced that this reaction barrier is readily surpassed under
the given reaction conditions. The next step of the reaction is
initiated by an intramolecular proton transfer via heterolytic
cleavage of the coordinated H2 which leads to a simultaneous
cleavage of one C−O bond and formation of an O−H bond.
As a result, water and Fe-bound CO are formed while BH2OH
is eliminated, resulting in complex G. Although the formation
of G is exothermic by ΔG(G/F) = −24.3 kcal mol−1, it is
associated with a barrier of ΔG⧧(TS-F-G) = 24.7 kcal mol−1
due to the breaking and formation of multiple bonds.
Analogous to E, G isomerizes to a Fe(0)−H2 species and
subsequently binds a second CO2 through the carbon to
generate H. Intermediate H in turn eliminates water with a
barrier of ΔG⧧(TS-H-13) = 39.0 kcal mol−1 to give the final
product 13, which has been characterized by X-ray
crystallography. As expected, 13 constitutes the lowest point
on the potential energy surface (ΔG(13/11) = −41.3 kcal
mol−1). To this end, we would like to point out that the high
barriers on the way to 13 can only be overcome because of the
applied reaction conditions that favor the thermodynamic
product and, to a lesser degree, the considerable gain in free
While the reduction of CO2 by 11 in acetonitrile is predicted
to proceed through “normal” insertion into B−H and Fe−H
bonds, our results indicate that the reaction in THF follows a
different route. The Triphos ligand scaffold offers steric
−
protection to the Fe(II) center, leaving only the BH4 moiety
chemically active. Our calculations indicate that cleaving one of
the Triphos−Fe bonds is highly energetically unfavorable
(Figure S40). Therefore, the Triphos does not actively
participate in the reaction. On the other hand, THF, being a
comparably bulky solvent, cannot bind to the Fe center in a
fashion similar to acetonitrile. Consequently, no reaction
between 11 and CO2 is observed at ambient temperature and
pressure. Only at an elevated temperature of 70 °C and a
pressure of 7 bar the final product 13 is immediately formed.
In contrast to the above-discussed reaction, no spectroscopic
evidence for stable intermediates was found, leaving only
theory-based arguments to elucidate the most probable
H
DOI: 10.1021/acs.organomet.8b00711
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
energy associated with some of the previous steps.
Furthermore, since the produced BH2OH is only metastable,
it is likely to further react with water, BH2OH, or other boron
containing degradation products. Even though the ultimate
boron containing degradation product could not yet be
characterized, the concomitant gain in free energy adds to
the driving force of the observed total reaction.
AP242/2-1, and Cluster of Excellence RESOLV EXC1069),
the Fraunhofer Internal Programs under Grant No.
ATTRACT 097-602175 as well as the Max Planck Gesellschaft
(M.v.G.). F.W. thanks the Studienstiftung des deutschen
Volkes for a Ph.D. fellowship. M.R. and A.K. gratefully
acknowledge funding by the Otto-Hahn program of the Max-
Planck Gesellschaft. The presented quantum chemical
calculations were partially conducted on computers provided
CONCLUSION
by the Zentrum fur molekulare Spektroskopie und Simulation
̈
solvensgesteuerter Prozesse (ZEMOS). We thank R. Miller
and M. Reback for support in language polishing.
■
In conclusion, a unique example of a solvent-controlled CO2
reduction is reported. While the reduction of CO2 is facilitated
by [(TriphosC)Fe(H)(BH4)] in stoichiometric conversions,
our results show that the solvent also plays a significant role in
the reduction pathway of CO2 and needs to be investigated. By
performing comparative spectroscopic and quantum chemical
studies, we provide reasonable pathways for the different
reactions observed. Here, the solvent coordination controls if a
Fe(0)−(H2) or a Fe(II)−(H−) intermediate is formed, leading
either to a C- or O-coordinated formate intermediate,
respectively. Thus, the reaction either leads to the formation
of CO or formic acid in a highly selective manner. These
results could clearly stimulate the further investigation of the
role of solvents in organometallic catalysis and provides
valuable information on the control of CO2 conversion.
ABBREVIATIONS
■
COT, cyclooctatetraen; DFT, density functional theory;
DLPNO-CCSD(T), coupled cluster theory with singles
doubles and perturbative triples in its domain-based pair
natural orbital implementation; CC-PVXZ (X = TQ),
correlation consistent polarized valence triple (T) or quadruple
(Q) zeta basis sets; def2-SVP, split-valence polarized basis set;
def2-TZVP, triple-ζ valence polarized basis set; PCM,
polarizable continuum model; TDDFT, time-dependent linear
response density functional theory
REFERENCES
■
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.organo-
met.8b00711.
■
(1) Qiao, J.; Liu, Y.; Hong, F.; Zhang, J. A Review of Catalysts for
the Electroreduction of Carbon Dioxide to Produce Low-Carbon
Fuels. Chem. Soc. Rev. 2014, 43 (2), 631−675.
S
(2) Francke, R.; Schille, B.; Roemelt, M. Homogeneously Catalyzed
Electroreduction of Carbon DioxideMethods, Mechanisms, and
Catalysts. Chem. Rev. 2018, 118 (9), 4631−4701.
Methods, additional calculations and experimental
details, and supportive spectroscopic data (PDF)
(3) Field, L. D.; Lawrenz, E. T.; Shaw, W. J.; Turner, P. Insertion of
CO 2CS 2 and COS into Iron(II)−Hydride Bonds. Inorg. Chem. 2000,
39 (25), 5632−5638.
Accession Codes
(4) Allen, O. R.; Dalgarno, S. J.; Field, L. D. Reductive
Disproportionation of Carbon Dioxide at an Iron(II) Center.
Organometallics 2008, 27 (14), 3328−3330.
CCDC 1871381−1871387 contain the supplementary crys-
tallographic 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.
(5) Yu, Y.; Sadique, A. R.; Smith, J. M.; Dugan, T. R.; Cowley, R. E.;
Brennessel, W. W.; Flaschenriem, C. J.; Bill, E.; Cundari, T. R.;
Holland, P. L. The Reactivity Patterns of Low-Coordinate Iron−
Hydride Complexes. J. Am. Chem. Soc. 2008, 130 (20), 6624−6638.
(6) Sadique, A. R.; Brennessel, W. W.; Holland, P. L. Reduction of
AUTHOR INFORMATION
Corresponding Authors
*E-mail: ulf.apfel@rub.de.
*E-mail: mroemelt@theochem.ruhr-uni-bochum.de.
CO to CO Using Low-Coordinate Iron: Formation of a Four-
■
2
Coordinate Iron Dicarbonyl Complex and a Bridging Carbonate
Complex. Inorg. Chem. 2008, 47 (3), 784−786.
(7) Langer, R.; Diskin-Posner, Y.; Leitus, G.; Shimon, L. J. W.; Ben-
David, Y.; Milstein, D. Low-Pressure Hydrogenation of Carbon
Dioxide Catalyzed by an Iron Pincer Complex Exhibiting Noble Metal
Activity. Angew. Chem., Int. Ed. 2011, 50 (42), 9948−9952.
ORCID
Maurice van Gastel: 0000-0002-1547-6365
Michael Roemelt: 0000-0002-4780-5354
Ulf-Peter Apfel: 0000-0002-1577-2420
̈
(8) Bertini, F.; Gorgas, N.; Stoger, B.; Peruzzini, M.; Veiros, L. F.;
Kirchner, K.; Gonsalvi, L. Efficient and Mild Carbon Dioxide
Hydrogenation to Formate Catalyzed by Fe(II) Hydrido Carbonyl
Complexes Bearing 2,6-(Diaminopyridyl)Diphosphine Pincer Li-
gands. ACS Catal. 2016, 6 (5), 2889−2893.
Author Contributions
The manuscript was written through contributions of all
authors. All authors have given approval to the final version of
the manuscript.
(9) Fong, H.; Peters, J. C. Hydricity of an Fe−H Species and
Catalytic CO Hydrogenation. Inorg. Chem. 2015, 54 (11), 5124−
2
Author Contributions
5135.
⊥L.I., A.K.: These authors contributed equally.
(10) Zall, C. M.; Linehan, J. C.; Appel, A. M. Triphosphine-Ligated
Copper Hydrides for CO 2 Hydrogenation: Structure, Reactivity, and
Thermodynamic Studies. J. Am. Chem. Soc. 2016, 138 (31), 9968−
9977.
Notes
The authors declare no competing financial interest.
(11) Federsel, C.; Boddien, A.; Jackstell, R.; Jennerjahn, R.; Dyson,
P. J.; Scopelliti, R.; Laurenczy, G.; Beller, M. A Well-Defined Iron
Catalyst for the Reduction of Bicarbonates and Carbon Dioxide to
Formates, Alkyl Formates, and Formamides. Angew. Chem., Int. Ed.
2010, 49 (50), 9777−9780.
ACKNOWLEDGMENTS
■
This work was supported by the Fonds der Chemischen
Industrie (Liebig grant to U.-P.A.) and through the Deutsche
Forschungsgemeinschaft (Emmy Noether grant to U.-P.A.,
I
DOI: 10.1021/acs.organomet.8b00711
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
(12) Petuker, A.; Merz, K.; Merten, C.; Apfel, U.-P. Controlled
Flexible Coordination in Tripodal Iron(II) Phosphane Complexes:
Effects on Reactivity. Inorg. Chem. 2016, 55 (3), 1183−1191.
(13) Ghilardi, C. A.; Innocenti, P.; Midollini, S.; Orlandini, A.
Synthesis, Characterization, and Structure of the Complex [FeH-
(H2BH2){CH3C(CH2PPh2)3}]. J. Chem. Soc., Dalton Trans. 1985,
No. 4, 605.
(14) Guilera, G.; McGrady, G. S.; Steed, J. W.; Burchell, R. P. L.;
Sirsch, P.; Deeming, A. J. Synthesis and Characterisation of
[(Triphos)Fe(CO)H2] and Its Protonation to a Dihydrogen
Complex via an Unconventional Hydrogen-Bonded Intermediate.
New J. Chem. 2008, 32 (9), 1573.
Isomer Shifts: Meta-GGA and Double-Hybrid Functionals. Inorg.
Chem. 2009, 48 (3), 784−785.
(33) Neese, F.; Hansen, A.; Liakos, D. G. Efficient and Accurate
Approximations to the Local Coupled Cluster Singles Doubles
Method Using a Truncated Pair Natural Orbital Basis. J. Chem. Phys.
2009, 131 (6), 064103.
(34) Riplinger, C.; Sandhoefer, B.; Hansen, A.; Neese, F. Natural
Triple Excitations in Local Coupled Cluster Calculations with Pair
Natural Orbitals. J. Chem. Phys. 2013, 139 (13), 134101.
(35) Riplinger, C.; Neese, F. An Efficient and near Linear Scaling
Pair Natural Orbital Based Local Coupled Cluster Method. J. Chem.
Phys. 2013, 138 (3), 034106.
(36) Riplinger, C.; Pinski, P.; Becker, U.; Valeev, E. F.; Neese, F.
Sparse MapsA Systematic Infrastructure for Reduced-Scaling
Electronic Structure Methods. II. Linear Scaling Domain Based Pair
Natural Orbital Coupled Cluster Theory. J. Chem. Phys. 2016, 144
(2), 024109.
(15) Herold, S.; Mezzetti, A.; Venanzi, L. M.; Albinati, A.; Lianza, F.;
Gerfin, T.; Gramlich, V. Synthetic Methodologies for Tripodal
Phosphines. The Preparation of MeSi(CH2PPh2)3 and n-BuSn-
(CH2PPh2)3 and a Comparison of Their Rhodium(I) and
Ruthenium(II) Coordination Chemistry. The X-Ray Crystal
Structures of [Rh(NBD){n-BuSn(CH2PPh2)3}](OTf) and [Rh-
(NBD){MeSi(CH2PPh2)3}](OTf). Inorg. Chim. Acta 1995, 235
(1−2), 215−231.
(16) Neese, F. The ORCA Program System: The ORCA Program
System. Wiley Interdiscip. Rev. Comput. Mol. Sci. 2012, 2 (1), 73−78.
(17) Neese, F. Software Update: The ORCA Program System,
Version 4.0: Software Update. Wiley Interdiscip. Rev. Comput. Mol. Sci.
2018, 8 (1), No. e1327.
(18) Becke, A. D. Density-Functional Exchange-Energy Approx-
imation with Correct Asymptotic Behavior. Phys. Rev. A: At., Mol.,
Opt. Phys. 1988, 38 (6), 3098−3100.
(19) Perdew, J. P. Density-Functional Approximation for the
Correlation Energy of the Inhomogeneous Electron Gas. Phys. Rev.
B: Condens. Matter Mater. Phys. 1986, 33 (12), 8822−8824.
(20) Weigend, F.; Ahlrichs, R. Balanced Basis Sets of Split Valence,
Triple Zeta Valence and Quadruple Zeta Valence Quality for H to Rn:
Design and Assessment of Accuracy. Phys. Chem. Chem. Phys. 2005, 7
(18), 3297.
(37) Liakos, D. G.; Neese, F. Improved Correlation Energy
Extrapolation Schemes Based on Local Pair Natural Orbital Methods.
J. Phys. Chem. A 2012, 116 (19), 4801−4816.
(38) Neese, F. ORCA Manual.
(39) Dunning, T. H. Gaussian Basis Sets for Use in Correlated
Molecular Calculations. I. The Atoms Boron through Neon and
Hydrogen. J. Chem. Phys. 1989, 90 (2), 1007−1023.
(40) Woon, D. E.; Dunning, T. H. Gaussian Basis Sets for Use in
Correlated Molecular Calculations. III. The Atoms Aluminum
through Argon. J. Chem. Phys. 1993, 98 (2), 1358−1371.
(41) Balabanov, N. B.; Peterson, K. A. Systematically Convergent
Basis Sets for Transition Metals. I. All-Electron Correlation
Consistent Basis Sets for the 3d Elements Sc−Zn. J. Chem. Phys.
2005, 123 (6), 064107.
(42) Balabanov, N. B.; Peterson, K. A. Basis Set Limit Electronic
Excitation Energies, Ionization Potentials, and Electron Affinities for
the 3d Transition Metal Atoms: Coupled Cluster and Multireference
Methods. J. Chem. Phys. 2006, 125 (7), 074110.
(43) Cramer, C. J. Essentials of Computational Chemistry: Theories
and Models, 2nd ed.; Wiley: Chichester, U.K.; Hoboken, NJ, 2004.
(44) Jensen, F. Introduction to Computational Chemistry, 3rd ed.;
John Wiley & Sons: Chichester, U.K.; Hoboken, NJ, 2017.
(45) Barone, V.; Cossi, M. Quantum Calculation of Molecular
Energies and Energy Gradients in Solution by a Conductor Solvent
Model. J. Phys. Chem. A 1998, 102 (11), 1995−2001.
(21) Becke, A. D. A New Mixing of Hartree−Fock and Local
Density functional Theories. J. Chem. Phys. 1993, 98 (2), 1372−1377.
(22) Becke, A. D. Density functional Thermochemistry. III. The
Role of Exact Exchange. J. Chem. Phys. 1993, 98 (7), 5648−5652.
(23) Dunlap, B. I.; Connolly, J. W. D.; Sabin, J. R. On Some
Approximations in Applications of X α Theory. J. Chem. Phys. 1979,
71 (8), 3396−3402.
̈
(46) Behrens, H.; Feilner, H.-D.; Lindner, E. Uber Reaktionen von
̈
(24) Vahtras, O.; Almlof, J.; Feyereisen, M. W. Integral
̈
Cyclooctatetraen-eisen-tricarbonyl mit dreizahnigen N- und P-
Approximations for LCAO-SCF Calculations. Chem. Phys. Lett.
1993, 213 (5−6), 514−518.
Liganden. Z. Anorg. Allg. Chem. 1971, 385 (3), 321−324.
(47) Ghilardi, C. A.; Innocenti, P.; Midollini, S.; Orlandini, A.
Synthesis and Molecular Structure of Hydrido(Tetrahydroborato)-
(1,1,1-Tris(Diphenylphosphinomethyl)Ethane)-Iron(II). J. Organo-
met. Chem. 1982, 231 (4), C78−C80.
(25) Neese, F.; Wennmohs, F.; Hansen, A.; Becker, U. Efficient,
Approximate and Parallel Hartree−Fock and Hybrid DFT Calcu-
lations. A ‘Chain-of-Spheres’ Algorithm for the Hartree−Fock
Exchange. Chem. Phys. 2009, 356 (1−3), 98−109.
(48) Dapporto, P.; Fallani, G.; Midollini, S.; Sacconi, I. Preparation,
Properties, and Structure of a Diamagnetic Dimeric Diiron
Compound with Three Bridge Hydrogen Bonds. J. Am. Chem. Soc.
1973, 95 (6), 2021−2023.
(49) Dapporto, P.; Midollini, S.; Sacconi, L. Synthesis, Character-
ization, and x-Ray Structure of Confacial-Bioctahedral Iron(II)- and
Cobalt(II)-Hydrido Complexes with Tridentate Tripod Ligands
Containing Phosphorus and Arsenic as the Donor Atoms. Inorg.
Chem. 1975, 14 (7), 1643−1650.
́
(26) Izsak, R.; Neese, F. An Overlap Fitted Chain of Spheres
Exchange Method. J. Chem. Phys. 2011, 135 (14), 144105.
́
(27) Izsak, R.; Neese, F.; Klopper, W. Robust Fitting Techniques in
the Chain of Spheres Approximation to the Fock Exchange: The Role
of the Complementary Space. J. Chem. Phys. 2013, 139 (9), 094111.
(28) Weigend, F. Accurate Coulomb-Fitting Basis Sets for H to Rn.
Phys. Chem. Chem. Phys. 2006, 8 (9), 1057.
(29) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and
Accurate Ab Initio Parametrization of Density Functional Dispersion
Correction (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys. 2010,
132 (15), 154104.
(50) Petuker, A.; Merten, C.; Apfel, U.-P. Modulating Sonogashira
Cross-Coupling Reactivity in Four-Coordinate Nickel Complexes by
Using Geometric Control: Modulating Cross-Coupling Reactivity in
Ni Complexes. Eur. J. Inorg. Chem. 2015, 2015 (12), 2139−2144.
(51) Chen, J.; Szalda, D. J.; Fujita, E.; Creutz, C. Iron(II) and
Ruthenium(II) Complexes Containing P, N, and H Ligands:
Structure, Spectroscopy, Electrochemistry, and Reactivity. Inorg.
Chem. 2010, 49 (20), 9380−9391.
(30) Grimme, S.; Ehrlich, S.; Goerigk, L. Effect of the Damping
Function in Dispersion Corrected Density Functional Theory. J.
Comput. Chem. 2011, 32 (7), 1456−1465.
(31) Neese, F. Prediction and Interpretation of the 57Fe Isomer
̈
Shift in Mossbauer Spectra by Density Functional Theory. Inorg.
Chim. Acta 2002, 337, 181−192.
̈
(32) Romelt, M.; Ye, S.; Neese, F. Calibration of Modern Density
(52) Taheri, A.; Thompson, E. J.; Fettinger, J. C.; Berben, L. A. An
57
̈
Functional Theory Methods for the Prediction of Fe Mossbauer
Iron Electrocatalyst for Selective Reduction of CO to Formate in
2
J
DOI: 10.1021/acs.organomet.8b00711
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Water: Including Thermochemical Insights. ACS Catal. 2015, 5 (12),
7140−7151.
(53) Tsay, C.; Livesay, B. N.; Ruelas, S.; Yang, J. Y. Solvation Effects
on Transition Metal Hydricity. J. Am. Chem. Soc. 2015, 137 (44),
14114−14121.
(54) Taheri, A.; Berben, L. A. Tailoring Electrocatalysts for Selective
+
CO or H Reduction: Iron Carbonyl Clusters as a Case Study.
2
Inorg. Chem. 2016, 55 (2), 378−385.
(55) Pitman, C. L.; Brereton, K. R.; Miller, A. J. M. Aqueous
Hydricity of Late Metal Catalysts as a Continuum Tuned by Ligands
and the Medium. J. Am. Chem. Soc. 2016, 138 (7), 2252−2260.
(56) Ceballos, B. M.; Tsay, C.; Yang, J. Y. CO 2 Reduction or HCO
−
Oxidation? Solvent-Dependent Thermochemistry of a Nickel
2
Hydride Complex. Chem. Commun. 2017, 53 (53), 7405−7408.
(57) Burgess, S. A.; Appel, A. M.; Linehan, J. C.; Wiedner, E. S.
Changing the Mechanism for CO Hydrogenation Using Solvent-
2
Dependent Thermodynamics. Angew. Chem., Int. Ed. 2017, 56 (47),
15002−15005.
(58) Zhang, Y.; MacIntosh, A. D.; Wong, J. L.; Bielinski, E. A.;
Williard, P. G.; Mercado, B. Q.; Hazari, N.; Bernskoetter, W. H. Iron
Catalyzed CO Hydrogenation to Formate Enhanced by Lewis Acid
2
Co-Catalysts. Chem. Sci. 2015, 6 (7), 4291−4299.
(59) Gennaro, A.; Isse, A. A.; Vianello, E. Solubility and
Electrochemical Determination of CO2 in Some Dipolar Aprotic
Solvents. J. Electroanal. Chem. Interfacial Electrochem. 1990, 289 (1−
2), 203−215.
(60) Knopf, I.; Cummins, C. C. Revisiting CO Reduction with
2
NaBH 4 under Aprotic Conditions: Synthesis and Characterization of
Sodium Triformatoborohydride. Organometallics 2015, 34 (9), 1601−
1603.
(61) Zhao, Y.; Zhang, Z. Thermodynamic Properties of CO
2
Conversion by Sodium Borohydride. Chem. Eng. Technol. 2015, 38
(1), 110−116.
(62) Collin, J. P.; Jouaiti, A.; Sauvage, J. P. Electrocatalytic
Properties of (Tetraazacyclotetradecane)Nickel(2+) and Ni2-
(Biscyclam)4+ with Respect to Carbon Dioxide and Water
Reduction. Inorg. Chem. 1988, 27 (11), 1986−1990.
(63) Song, J.; Klein, E. L.; Neese, F.; Ye, S. The Mechanism of
Homogeneous CO Reduction by Ni(Cyclam): Product Selectivity,
2
Concerted Proton−Electron Transfer and C−O Bond Cleavage.
Inorg. Chem. 2014, 53 (14), 7500−7507.
́
(64) Rosas-Hernandez, A.; Junge, H.; Beller, M.; Roemelt, M.;
Francke, R. Cyclopentadienone Iron Complexes as Efficient and
Selective Catalysts for the Electroreduction of CO 2 to CO. Catal. Sci.
Technol. 2017, 7 (2), 459−465.
K
DOI: 10.1021/acs.organomet.8b00711
Organometallics XXXX, XXX, XXX−XXX