Diferrate [Fe2(CO)6(μ-CO){μ-P(aryl)2}]? as Self-Assembling Iron/Phosphor-Based Catalyst for the Hydrogen Evolution Reaction in Photocatalytic Proton Reduction—Spectroscopic Insights

Article abstract of DOI:10.1002/chem.201802694

This work is focused on the identification and investigation of the catalytically relevant key iron species in a photocatalytic proton reduction system described by Beller and co-workers. The system is driven by visible light and consists of the low-cost [Fe₃(CO)₁₂] as catalyst precursor, electron-poor phosphines P(R)₃ as co-catalysts, and a standard iridium-based photosensitizer dissolved in a mixture of THF, water, and the sacrificial reagent triethylamine. The catalytic reaction system was investigated by operando continuous-flow FTIR spectroscopy coupled with H₂ gas volumetry, as well as by X-ray absorption spectroscopy, NMR spectroscopy, DFT calculations, and cyclic voltammetry. Several iron carbonyl species were identified, all of which emerge throughout the catalytic process. Depending on the applied P(R)₃, the iron carbonyl species were finally converted into [Fe₂(CO)₆(μ-CO){μ-P(R)₂}]^?. This involves a P?C cleavage reaction. The requirements of P(R)₃ and the necessary reaction conditions are specified. [Fe₂(CO)₆(μ-CO){μ-P(R)₂}]^? represents a self-assembling, sulfur-free [FeFe]-hydrogenase active-site mimic and shows good catalytic activity if the substituent R is electron poor. Deactivation mechanisms have also been investigated, for example, the decomposition of the photosensitizer or processes observed in the case of excessive amounts of P(R)₃. [Fe₂(CO)₆(μ-CO){μ-P(R)₂}]^? has potential for future applications.

Full text of DOI:10.1002/chem.201802694

A Journal of
Accepted Article
Title: Diferrate [Fe2(CO)6(ꢀ-CO)(ꢀ-P(aryl)2)]− as Self Assembling
Iron/Phosphor Based Catalyst for Hydrogen Evolution in
Photocatalytic Proton Reduction - Spectroscopic Insights
Authors: Ralf Ludwig, Steffen Fischer, Arend Rösel, Anja Kammer,
Enrico Barsch, Roland Schoch, Henrik Junge, Matthias
Bauer, and Matthias Beller
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201802694
Link to VoR: http://dx.doi.org/10.1002/chem.201802694
Supported by

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
Diferrate [Fe2(CO)6(-CO)(-P(aryl)2)]− as Self Assembling Iron/
Phosphor Based Catalyst for the Hydrogen Evolution Reaction in
Photocatalytic Proton Reduction – Spectroscopic Insights
Steffen Fischer,[a,b] Arend Rösel,[a] Anja Kammer,[c] Enrico Barsch,[a] Roland Schoch,[d] Henrik Junge,[c]
Matthias Bauer,[d] Matthias Beller[c] and Ralf Ludwig*[a,b,c]
Abstract: This work focuses on the identification and investigation of
catalytically relevant iron key species in a photocatalytic proton
reduction system described by Beller and co-workers. The system is
driven by visible light and consists of low cost [Fe3(CO)12] as catalyst
precursor, electron poor phosphines P(R)3 as co-catalyst, and a
requires the application of a sacrificial reagent (SR) as electron
donor. Furthermore, a light harvesting and charge separating
photosensitizer (PS) is necessary. The PS transfers electrons to
a catalyst which combines them with protons from water to
accomplish the Hydrogen Evolution Reaction: the proton
standard iridium based photosensitizer dissolved in a mixture of THF, reduction catalyst (catalyst, also named: water reduction
water and the sacrificial reagent triethylamine. Operando continuous
flow FTIR spectroscopy coupled with H2 gas volumetry is applied, as
well as XAS, NMR spectroscopy, DFT calculations and cyclic
voltammetry. Several iron carbonyl species are identified, all of
which emerge throughout the process. Depending on the applied
P(R)3, the iron carbonyl species are finally converted into
[Fe2(CO)6(-CO)(-P(R)2)]−. Requirements on P(R)3 and necessary
reaction conditions are specified. Fe2(CO)6(-CO)(-P(R)2)]−
represents a self assembling, sulfur free [FeFe]-hydrogenase active
site mimic and shows good catalytic activity if the substituent R is
electron poor. Deactivation mechanisms are also investigated, e.g.
decomposition of the photosensitizer or processes observed in case
of excessive amounts of P(R)3. [Fe2(CO)6(-CO)(-P(R)2)]− is
potentially relevant for future applications.
catalyst).[3] The latter is preferably made of abundant and low
cost metals, such as Fe, Co, Ni, Cu or Zn and must show a long
term stability.[4] Molecular iron based catalysts have mainly been
inspired by the structure of the active site of [FeFe]-hydrogenase
([FeFe]-H2ase)[5] (Scheme 1, grey background), which has been
in focus of numerous reviews.[2b, 4d, 6]
Introduction
The use of sunlight, more precisely, the storage of solar energy
in form of chemical energy carriers represents one of the most
promising options to approach current energy challenges.[1]
Photocatalytic water splitting into oxygen and energy containing
hydrogen is very suitable for this purpose. Its two half reactions
water oxidation and water reduction are mostly studied
separately. This allows for simplification of the system and
facilitates a detailed understanding, which is beneficial for further
improvements.[2] The photocatalytic water reduction reaction
[a]
[b]
S. Fischer, A. Rösel, E. Barsch, Prof. Dr. R. Ludwig
Physical and Theoretical Chemistry Department
University of Rostock
Dr.-Lorenz-Weg 2, 18059 Rostock (Germany)
E-mail: ralf.ludwig@uni-rostock.de
S. Fischer, Prof. Dr. R. Ludwig
Department Life, Light & Matter
University of Rostock
Albert-Einstein-Straße 25, 18059 Rostock (Germany)
A. Kammer, Dr. H. Junge, Prof. Dr. M. Beller
Leibniz-Institut für Katalyse e. V. (LIKAT Rostock)
Albert-Einstein-Straße 29a, 18059 Rostock (Germany)
R. Schoch, Prof. Dr. M. Bauer
[c]
[d]
Department Chemie, Fakultät Naturwissenschaften
Paderborn University
Warburger Str. 100, 33098 Paderborn (Germany)
Scheme 1. [FeFe]-H2ase active site structure and its mimics as catalysts in
the photocatalytic proton reduction.
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
colleagues[14] revealed important mechanistic information, which
will be summarized within the following chapters.
The Beller system was later modified by the addition of specific
electron-poor aryl phosphines (P(R)3) as co-catalysts (Scheme 1,
bottom).[14a] Namely, the addition of P(C6H4-4-CF3)3 (P(d)3) leads
to an increased TON by approximately 50 % up to 675 after 3
hours. Due to optimized reaction conditions and application of
P(C6H4-3,5-(CF3)2)3 (P(c)3) as co-catalyst, a TON of even 1610
was achieved after 24 hours. Notably, the reaction can be driven
with visible light only. This was shown by the application of 385
and 420 nm UV cut-off filters, which led to a decrease of the
TON by not more than respectively 5 and 10 % as compared to
the measurements with unfiltered Xe-light. Further improvement
of the system was accomplished by usage of an Ir-PS derivate
with an isopropyl substituent (Ir-PSiPr, Scheme 2) leading to a
final TON of 2770 after 20 hours of irradiation with blue light
(440 nm).[15]
Acknowledging these high TON’s and taking into account the
widely differing reaction scale or photosensitizer applied in the
aforementioned systems, 3CO together with P(R)3 is still
representing an outstanding iron catalyst by now, which
combines an easy access with a comparably high performance.
Du and Eisenberg highlighted the system of Beller and co-
workers in a review while at the same time remarking the
obscurity of the phosphine’s role.[4a] Moreover, it had not been
clear, why phosphites or phosphines that differ from the above-
mentioned do not enhance or even reduce the catalyst´s
performance. Another unexplained issue had been the decline in
activity which is observed if the otherwise enhancing P(d)3 is
applied at amounts higher than one equivalent respective to
3CO.[14a]
In this work, we elucidate these issues by monitoring the system
by operando continuous flow FTIR spectroscopy coupled with
simultaneous measurement of the evolving H2 volume. We
therefore applied various types of P(R)3 with different electron
acceptor/donor properties, starting with electron poor P(OPh)3
(P(a)3) through to the electron rich alkyl phosphine P(Bu)3 (P(i)3)
(Table 1, first column). For P(d)3 we additionally performed a
more detailed analysis by NMR spectroscopy, DFT calculation
methods and cyclic voltammetry. Furthermore, X-ray absorption
spectroscopy (XAS) was carried out, to substantiate our analysis.
We start the following section by describing the species that
emerge throughout the reaction if 3CO and one equivalent
phosphine are applied. Subsequently, we present the results of
our XAS study and specify the catalytic relevance of the
identified species. At the end of this work, we point out
deactivation processes of the system and illustrate our results of
electrochemistry experiments.
Scheme 2. Photosensitizers in the photocatalytic proton reduction.
Embedded in a protein matrix, the diiron complex shows specific
structure motives e.g. terminal cyanide and carbonyl ligands as
well as a bridging carbonyl and an azadithiolato ligand (adtH).
Whereby the nitrogen bridgehead in adtH acts as pendant base
and plays an important role in the biological catalytic cycle.
A successful [FeFe]-H2ase active site mimic applied in a
photocatalytic water reduction reaction was introduced by Wang,
Sun and co-workers.[7] They used [Fe2(adtBz)(CO)5(P(Pyr)3)]
(Scheme 1, Pyr = N-pyrrolyl) as catalyst together with an iridium
based PS [Ir(bpy)(ppy)2]PF6 (Ir-PS, Scheme 2) and triethylamine
(TEA) as SR. The diiron complex achieved a turn over number
(TON) of 466 molecules H2 per catalyst molecule after 8 hours of
light irradiation.
Zeng, Yang, Li and co-workers applied the same PS and SR.
Concerning the catalyst, they made use of dendritic
architectures that encapsulate an active site model of the
[FeFe]-H2ase to mimic the protein matrix of the enzyme.[8] From
one to four dendrimer generations (e.g. two generations
depicted in Scheme 1) TONs of 18 100 to 22 200 were observed.
However, the system was only tested at a very small scale with
an amount of 10 nM catalyst. Apart from that, systems like these
show that the nitrogen atom in the thiolatebridge of the catalyst
is not indispensable, since abiological reaction mechanisms are
also suitable for reduction of protons.[6e]
With [FeFe]-H2ase active site mimics in combination with a
different class of PS, namely quantum dots like CdTe, TONs of
up to 52 800 have been observed by Jian and Wu et al..[9] The
same group revealed that this high TON should mainly be
attributed to the PS rather than the catalyst.[10] In fact, a similar
system using [Fe2(-S(CH2)2SO3Na)2(CO)6] as catalyst (Scheme
1) and ascorbic acid as SR shows a TON of 26 500 if the
quantum dot CdSe (Scheme 2) was applied, but only a TON of
178 if a ruthenium based PS [Ru(bpy)3][PF6]2 (Ru-PS, Scheme
2) was used. (The latter, belongs to the same class as the Ir-
PS)[11][12]
A thiolate free photocatalytic system with a self assembling
catalyst was presented by Beller and co-workers.[13] This system
consists of the Ir-PS and commercially available iron carbonyls
such as [Fe3(CO)12] (3CO) as catalyst precursor. Thus,
time-consuming synthesis of the catalyst is avoided. The
Results and Discussion
Emerging Species due to Application of 3CO and one
equivalent Phosphine: In the original phosphine free system of
Beller and co-workers, a preliminary reaction preceding to light
irradiation is observed: The moment the precursor [Fe3(CO)12]
(3CO) is dissolved in THF/TEA/H2O, it starts to convert within
components are dissolved in
a
THF/TEA/H2O mixture
(volumetric ratio of 4/1/1), whereas TEA again acts as SR. After
6 h of irradiation by visible light, a TON of 400 was achieved.
Extensive spectroscopic studies by our working group and
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
two minutes into a mixture of [Fe3(CO)11]•−, [Fe2(CO)8]•−, and
[Fe(CO)5]. This is caused by the presence of TEA and water and
was reported in our previous publications.[14b, 14d]
direct comparison with literature see Table SI1 in the Supporting
Information).
If the system is modified by the additional application of one of
the phosphines P(b)3, P(d)3, P(e)3, P(f)3, P(h)3 or P(i)3 (Table 1),
the same conversion reaction of 3CO takes place, but
Table 1. Cone angles[a] of phosphines[b] () and IR data[c] of in-situ generated
iron phosphides (ṽ).
−
additionally
a
further species emerges, identified as
R
P(R)3
1P(R)3:
2COP(R)2
:
 / °
ṽ (t-CO) / cm−1
ṽ (t-CO) / cm−1
[Fe(CO)4(P(R)3)] (1P(R)3, Figure SI1 in the Supporting
Information). The assignment of 1P(R)3 is supported by 31P-
NMR analysis of a solution of 3CO with one equivalent P(d)3 in
THF/TEA/H2O (ratio 4/1/1). The spectrum shows a singlet at 
76.1 ppm, which is in good agreement with the signal for 1P(d)3
reported by Howell et al.[16] to be at  78.8 ppm in CDCl3.[17]
Noteworthy, the phosphine ligand coordinates quantitatively to
iron, since no signal is found for free P(d)3 at  -5.6 ppm
(Figure SI2 in the Supporting Information).
The conversion of 3CO and P(R)3 into 1P(R)3 has already been
reported to occur at significantly harsher conditions.[18] However,
in these reports it was proposed to be initiated by CO loss from
3CO[18c] and the formation of radicals[19], as it is also the case in
the reaction presented here ( vide supra).
a
184
n.f.
n.f.
b
c
128
2064 (m)
1989 (m)
1958 (s)
n.f.
160[d]
n.f.
2025 (m)
1977 (s)
1947 (m)
1926 (m)
d
e
f
145 or
164[d]
2054 (m)
1981 (m)
1947 (s)
2021 (m)
1971 (s)
1941 (m)
1921 (m)
1P(R)3 is not formed in case P(a)3, P(c)3 or P(g)3 is applied. This
might be explained by steric hindrance due to the phosphine’s
wide cone angles (Table 1, second column). While this
argument holds for P(a)3 and P(g)3, this is not necessarily the
case for P(c)3, since the latter has a cone angle similar to those
145 or
155[d]
2048 (m)
1971 (m)
1942 (s)
2016 (m)
1965 (s)
1934 (m)
1916 (m)
of P(d)3 and P(e)3 as reported by Howell et al.[16]
.
145
2045 (m)
1967 (m)
1938 (s)
2014 (m)
1963 (s)
1932 (m)
1913 (m)
In the next step of the catalytic system, hydrogen generation is
started by addition of the Ir-PS and initiation of light irradiation[20]
Our previous studies reveal that the Ir-PS is excited by light and
reductively quenched by TEA. The decomposition of the latter
gives rise to acetaldehyde, which shows an IR contribution at
1724 cm−1 (Scheme SI2 in the Supporting Information).[14a-c]
Subsequently, the iron carbonyls [Fe3(CO)11]•−, [Fe2(CO)8]•−, and
[Fe(CO)5] are converted into the hydride species [HFe3(CO11)]−
(H3−) and [HFe(CO)4]− (H1−) within one minute, due to electron
transfer from the reduced Ir-PS and reaction with water.[14a, 14d]
This happens in the phosphine free as well as the phosphine
containing system as proven by IR spectroscopy. The resulting
spectra (Figure 1 left, a and b, spectra at t = 2 min) show
contributions of H3− at 2064 (vw), 1999 (s), 1993 (s), 1975 (m),
1953 (w), and 1941 cm−1 (w) (Figure 1 middle, red spectrum)
together with contributions of H1− at 1998 (w), 1908 (m), and
1878 cm−1 (s) (Figure 1 middle, blue spectrum).
.
g
h
i
212
118
132
n.f.
n.f.
n.f.
n.f.
2044 (m)
1964 (m)
1931 (s)
2043 (m)
1964 (m)
1930 (s)
[a] cone angles reported by Tolman[23] if not otherwise stated. [b] the
substituents (R) of the phosphines are listed according to their electron
withdrawing character which decreases from top to bottom of the table. [c]
solvent: THF/TEA/H2O 4/1/1. [d] cone angles reported by Howell et al.[16]. t-CO
= terminal Carbonyls, n.f. = not formed, w = weak, m = medium, s = strong
Since 1P(R)3 remains in the solution during the beginning of the
reaction course of the modified system, its pure IR spectrum
could be revealed by subtracting the pure spectra of the known
compounds from the sum spectrum. The resulting spectrum of
the observed 1P(R)3 species (e.g. 1P(d)3, Figure 1 middle,
green spectrum) consists of three bands and shows the IR
profile of the well reported 1P(e)3 [21]. This profile is shifted to
higher or lower wavenumbers depending on the electron
acceptor/donor ability of the respective substituent (R) in the
applied phosphine. The wavenumbers are listed in Table 1 and
are in good agreement with literature values[16, 18a, 19, 21c, 22] (for
Only in case of application of P(c)3, P(d)3, P(e)3 and P(f)3, all
iron material is further converted into another iron phosphide
species,
which
was
identified
as
the
diferrate
[Fe2(CO)6(-CO)(-P(R)2)]− (2COP(R)2 ) and contains one
bridging and six terminal CO ligands as well as one bridging
phosphido ligand (Figure 1, right, violet labeled structure). The
complex shows four characteristic IR bands, which emerge
during the catalytic process in the spectral region of the terminal
carbonyl groups (e.g. R=d Figure 1, left, b, spectrum at t = 20 h).
The contribution of the bridging carbonyl ligand cannot be
−
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
Figure 1. Left: In-situ IR spectra of the reaction solution of a) the phosphine free system with 3CO as catalyst precursor, and b) the phosphine containing system
with 3CO and one equivalent of co-catalyst P(d)3 as well as c) with 3CO and three equivalents of P(d)3. For all three cases, two spectra are shown recorded at 2
min and 20 h after initiation of light irradiation. Reaction conditions: 0.5 mM Ir-PS, 0.305 mM 3CO, 20 mL THF/TEA/H2O (ratio: 4/1/1, pH 12), visible light (380-700
nm, 1.5 W), 25 °C, a) 0 mM P(d)3, b) 0.305 mM P(d)3, c) 0.915 mM P(d)3.
−
Middle: Pure component spectra of [HFe3(CO11)]− (H3−, red), [HFe(CO4)]− (H1−, blue), [Fe(CO)4(P(d)3)] (1P(d)3, green), [Fe2(CO)6(-CO)(-P(d)2)]− (2COP(d)2
,
violet) and [Fe2(CO)5(P(d)3)(-CO)(-P(d)2)]− (2P(d)3P(d)2−, turquoise). Spectra of H3−, H1−and 2COP(d)2 were recorded by measuring the respective
synthesized compound dissolved in THF/TEA/H2O (ratio: 4/1/1, pH 12). Spectra of 1P(d)3 and 2P(d)3P(d)2 were extracted by subtraction of the known
−
−
compounds from the in-situ spectra. Contributions of all five species in the in-situ IR spectra of the left figure are marked with * in the corresponding color.
Right: Conversion of the precursor 3CO + P(R)3 as detected by IR spectroscopy throughout the reaction. The scheme starts with processes observed in the
absence of light (grey background) and continues with the conversion of the phosphine deficient species occurring within minutes due to visible light irradiation.
−
Depending on the rest R, on a time scale of hours, all species are further converted into 2COP(R)2−. If an excess of P(R)3 is applied, formation of 2P(R)3P(R)2
occurs.
detected since it overlaps with the broad and intense band of
acetaldehyde, the decomposition product of TEA.
The observed wavenumbers of the bands for each 2COP(R)2
[NEt4][2COP(d)2], [NEt4][2COP(e)2] and [NEt4][2COP(f)2]. The IR
spectra of these synthesized compounds (e.g. [NEt4][2COP(d)2]
Figure 1, middle, violet) equal the corresponding in-situ spectra
observed in the catalytic process of the phosphine containing
system. Furthermore, they include the band of the bridging
carbonyl, which is otherwise hidden by acetaldehyde[14b] (Figure
SI10 in the Supporting Information).
−
complex are listed in Table 1. Again, they are shifted depending
−
on the electron acceptor/donor ability of R. For 2COP(e)2 , the
experimental wavenumbers of this work match with the IR data
24]
−
in the literature.[22d,
2COP(e)2 was first described by W.T.
Osterloh[25] and is the only one of the 2COP(R)2 compounds
reported here, which had been characterized[26] prior to this work
(Table SI2 in the Supporting Information). We further confirmed
Formation of 2COP(R)2 was also confirmed by NMR
−
−
spectroscopy. Therefore, a sample was taken from the solution
of the precursor system 3CO + P(d)3 after 15 hours of light
irradiation. At this time, all iron material is converted towards
−
the assignment of 2COP(R)2 by DFT calculation of the IR
−
spectra (Figure SI12 in the Supporting Information) and
synthesis of the preformed complexes [NEt4][2COP(c)2],
2COP(d)2 as reflected in the IR spectra. Consistently, the
31P-NMR spectrum shows only one singlet signal at  123,0 ppm,
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
which corresponds to the signal of synthesized [NEt4][2COP(d)2]
in THF-d8. (Figure SI2 and SI3 in the Supporting Information). In
Results of the analysis by X-ray absorption spectroscopy:
In order to further substantiate the results of IR analysis of the
formed species, XAS was carried out. Unlike IR spectroscopy,
XAS provides directly bond distances and coordinating atoms in
an element specific manner by evaluation of the extended X-ray
absorption fine structure (EXAFS) above the absorption edge.
However, the analysis of EXAFS spectra is complicated by the
fact that the structural parameters of all species in the reaction
solution are averaged in the spectral signal.[35]
−
the 19F-NMR spectrum, 2COP(d)2 appears at  −62.8 ppm. An
additional signal at
 −62.7 ppm could be assigned to
trifluorotoluene (dH) (Figure SI4 in the Supporting Information).
The latter derives from the cleavage of one of the substituents
from phosphorus during the formation of the phosphido bridge in
2COP(d)2− (P-C cleavage reaction, vide infra).
−
To determine the conditions for the formation of 2COP(R)2 , we
carried out experiments successively excluding one of the
system´s components apart from 3CO and P(R)3 with R = d.
They show that the Ir-PS is not necessary for the formation of
−
2COP(d)2 . On the other hand, the experiments indicate that all
other components like THF, TEA, H2O as well as light irradiation
are essential (Figure SI14 in the Supporting Information). The
need for TEA and H2O is most probably attributable to the fact
that they initiate the formation of radicals (vide supra).[14d] The
latter were found by Rahaman et al. to promote the generation
of a phosphido bridged diiron complex akin to 2COP(R)2 by
treatment of 3CO with Na+[Ph2CO] •− and P(2-furyl)3 in THF at
room temperature.[27]
The choice of the ligand P(R)3 was also found to be relevant: As
−
concluded from our study, the formation of 2COP(R)2 is only
observed if the phosphine meets the following two requirements:
(i) P(R)3 has to belong to the class of arylphosphines and (ii) the
ortho positions of the aryl group have to be unsubstituted.
Figure 2. Fourier transformed EXAFS function of the samples given in Table 2.
The Experimental spectrum is shown as black solid line while the fitted one is
given as grey dashed line.
These requirements are most probably attributable to the P-C or
P-O cleavage reaction, prior to the formation of the phosphido
bridge. Such transition metal mediated P-C cleavage reactions
have broadly been reported in literature.[28] They occur even at
28d]
mild conditions and temperatures as low as 25 °C.[28a,
Figure 2 shows the experimental Fourier transformed EXAFS
spectra, the according (k) functions and XANES (X-ray
absorption near edge structure) spectra are given in the
supplementary information. Analysis was carried out on Fourier
filtered data in the range of 1-3.2 Å using the radial distribution
function approach, because the nature of the formed species
was unknown. Since also mixtures of species were expected, no
multiple scattering analysis[36] was carried out and the rather
intense multiple scattering signal of the CO ligands was fitted
using a single scattering Fe-O shell[37]. The signal in case of
3CO+P(d)3+Ir-PS+h is slightly reduced compared to the other
spectra, as the fluorescence signal was superior to the
absorption data due to the presence of the Iridium
photosensitizer. However, the signal is affected by self
absorption effects. They are accounted for by an overall
amplitude reduction factor of 0.68, while for all others it was 0.9.
All obtained structural parameters are summarized in Table 2.
The parameters obtained for 3CO in THF/TEA/H2O (4/1/1)
solution agree very well with the previous EXAFS studies.[38] The
Fe-Fe coordination number deviates slightly from the expected
number of two, which is in line with the presence of iron species
of smaller nuclearity. Nonetheless the bond distances agree
very well with the crystal structure data of 3CO.[39]
Consistent with our observations, they commonly include
hydrido intermediates and yield bi- or higher-nuclearity
complexes featuring bridging phosphido groups.[28] The cleavage
[28a]
2
3
reactivity follows the order P-Csp > P-Csp > P-Csp
,
−
explaining the formation of 2COP(R)2 in case of application of
arylphosphines rather than alkylphosphines. For phosphites,
analogous P-O cleavage reactions with subsequent bridging
ligand formation are also known.[29] However, in-situ formation of
−
2COP(R)2 by application of phosphites is inhibited, since the
presence of water suppresses these P-O cleavage reactions[29a]
and can also induce decomposition of the ligands due to
hydrolysis[30]
.
The mechanism of the P-C cleavage reaction is not completely
investigated yet.[28c] So far, mechanistic studies point to the
oxidative addition of the P-C bond to the metal center as
28b, 31]
activating step.[28a,
This requires a coordinatively
unsaturated metal atom[32], e.g. provided by light initiated CO
dissociation. It is assumed that other processes precede the P-C
cleavage reaction. These processes include e.g. -coordination
of the phosphine´s aryl rings to neighbor metal atoms in
polynuclear complexes[33] or the oxidative addition of the ortho-
C-H bond to the metal center, known as “orthometallation”[28a, 28b]
These processes together with steric drawbacks might explain
why ortho substituted phosphines do not undergo P-C
.
By adding one equivalent P(d)3, a characteristic Fe-P shell
appears with a coordination number of 0.4. Together with the
significantly reduced Fe-Fe coordination number of 0.6, it is in
very good agreement with a quantitative coordination of P(d)3 to
iron (expected Fe-P coordination number 0.33). The average of
−
cleavage[34] and no formation of 2COP(R)2 is observed in case
of application of P(a)3 and P(g)3.
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
0.6 iron neighbors cannot be explained by only trinuclear
phosphine-free species, as in this case an average value >1
would be expected. Thus a mixture of mono-, di- and trinuclear
iron carbonyl compounds ([Fe(CO)5], [Fe2(CO)8]•−, and
[Fe3(CO)11]•−) is confirmed, which is also in agreement with the
Fe-C coordination number of 5.2.[40]
recognizable from the large Debye-Waller factor and is
confirmed by an analogous in-situ IR experiment carried out at
the same reaction conditions as the XAS experiment.
Accordingly, the IR spectrum measured at the same time shows
H3−, H1− and 1P(d)3 to be present besides the major species
−
2COP(d)2 with the biggest share of 43 % (Figure SI18 in the
Supporting Information).
Influence of 1P(R)3 and 2COP(R)2− on the H2 evolution: For a
better understanding of the differing effects of specific phosphine
co-catalysts, we start this section with a brief summary of the
results of previous mechanistic studies on the phosphine free
system with 3CO as exclusive precursor:
The catalytic cycle begins with the excitation of the Ir-PS by light
and the subsequent reductive quenching by TEA. The latter
process was found by means of photoluminescence
Table 2. Structural parameters obtained by fitting the experimental EXAFS
function with theoretical models.
Abs-Bs[e] N(Bs)[f]
R(Abs-Bs)[g]
[Å]

[h] [Å]
R[i] [%]
3CO[a]
3CO+
Fe-C
Fe-Fe
Fe-O[j]
4.7 ± 0.5
1.8 ± 0.3
7.0 ± 1.4
1.80 ± 0.02 0.067 ± 0.002 20.91
2.51 ± 0.03 0.032 ± 0.001
2.99 ± 0.03 0.067 ± 0.002
Fe-C
Fe-P
Fe-Fe
Fe-O[j]
5.2 ± 0.5
0.4 ± 0.1
0.6 ± 0.2
15.0 ± 3.2
1.75 ± 0.02 0.074 ± 0.002 16.88
2.23 ± 0.02 0.039 ± 0.001
2.71 ± 0.03 0.097 ± 0.020
2.99 ± 0.03 0.077 ± 0.002
spectroscopy and kinetic analysis to be the rate limiting step of
[b]
P(d)3
14d]
the system.[14c,
After that, the reduced Ir-PS transfers
electrons to the active catalyst, which we proposed to be H3−.
This proposition is based on time-dependent DFT calculations
and is supported by the finding, that H3− shows a more positive
3CO+
P(d)3+
Ir-PS
+h[c]
Fe-C
Fe-P
Fe-Fe
Fe-O[j]
3.5 ± 0.4
0.8 ± 0.2
1.6 ± 0.4
9.1 ± 1.8
1.78 ± 0.02 0.102 ± 0.020 18.22
2.13 ± 0.02 0.045 ± 0.009
2.46 ± 0.03 0.110 ± 0.022
2.94 ± 0.03 0.074 ± 0.015
14d]
reduction potential than the Ir-PS.[14a,
The transferred
electrons either are directly used within the catalytic cycle for
proton reduction and H2 generation or they induce fragmentation
of H3−. Due to the latter, quasi full conversion of H3− into the
resting state H1− occurs within one hour. However, both species
are in equilibrium since a light induced back reaction continually
regenerates H3− in small amounts (Scheme 3).[14d] Over the
further course of the reaction, H1− decomposes due to
processes discussed in a following chapter of this article (Figure
3, top blue concentration curve). After 16 hours, H1− is fully
consumed. As a consequence, catalysis ceases and gas
evolution stops (Figure 3, top grey gas evolution curve).[14d]
[NEt4]
[2COP(e)2][d] Fe-P
Fe-Fe
Fe-C
4[k]
1.75 ± 0.02 0.059 ± 0.012 22.73
2.16 ± 0.02 0.071 ± 0.002
2.53 ± 0.03 0.107 ± 0.022
2.99 ± 0.03 0.077 ± 0.016
1[k]
1[k]
Fe-O[j]
15 ± 3.0
[a]0.01 M 3CO in THF/TEA/H2O (4/1/1). [b]0.025 M 3CO and 0.025 M P(d)3 in
THF/TEA/H2O (ratio: 4/1/1, pH 12) measured 3 hours after preparation of the
reaction solution. [c]0.0033 M 3CO, 0.0033 M P(d)3 and 0.0022 M Ir-PS + h
in THF/TEA/H2O (ratio 4/1/1, pH 12), after 8 hours of reaction time. [d]0.05 M
[NEt4][2COP(e)2] in THF/TEA/H2O (ratio: 4/1/1, pH 12). [e]Abs= X-ray
absorbing atom, Bs=Backscattering atom. [f]Number of backscattering atoms.
[g]Distance between absorbing and backscattering atom. [h]Debye-Waller
factor. [i]Quality of fit. [j]Although denoted as a single Fe-O scattering path,
the underlying signal contains a significant amount of multiple scattering from
the CO ligand. [k]Value fixed to crystallographic number.
The system under working conditions (3CO+P(d)3+Ir-PS+h)
was subjected to a full EXAFS characterization after 8 hours of
reaction time. It is obvious that the nearest neighbor Fe-C
coordination number is reduced to around four, while the Fe-P
coordination number increases to nearly one. These
coordination numbers and the distances of both shells agree
very well within the error bar with those of the reference
compound [NEt4][2COP(e)2] with one coordinating bridging
phosphine and four CO ligands per iron center.[24b] The results
thus confirm the formation of 2COP(d)2− previously concluded by
IR analysis. However, due to the reduced reaction time and the
high concentrations of the reaction components (concentrations
are increased by a factor of 10 as compared to the other
experiments of this work), full conversion to this species is not
yet completed, which is obvious from the Fe-Fe coordination of
1.6 that is slightly too high even if the error is taken into account.
The presence of other species contributing to the spectra is also
Scheme 3. Equilibrium between H3− and H1−.
Modification of the system by application of one equivalent
phosphine ligand has varying effects on the H2 production,
depending on the choice of the phosphine ligand and the
emerging iron phosphides associated therewith.
As described previously, sterically demanding P(a)3 or P(g)3 do
not show any interaction with the iron species (Figure SI20a and
SI20g in the Supporting Information). Hence, their application
has no impact on the catalytic process and the same amount of
gas is produced as in the phosphine free system (Figure 4).
The phosphite P(b)3 or the alkylphosphines P(h)3 and P(i)3 do
coordinate to iron and form exclusively 1P(b)3, 1P(h)3 or 1P(i)3,
respectively. This has a small negative effect on the H2
productivity of the system most notably in case of the
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
alkylphosphines with a decline of the H2 yield by 15 % (Figure 4). even after decomposition of the latter. This displays the catalytic
−
−
Due to formation of 1P(R)3, less iron material is available for
activity of 2COP(d)2 . Hence, the formation of 2COP(d)2
transformation into the catalytically relevant species H3− and H1−. improves the stability of the system and increases the H2 yield
As a consequence, the latter are present in lower amounts and
are used up sooner over the reaction time. Since 1P(R)3 does
not contribute to the H2 production, gas evolution consequently
stops at an earlier stage as compared to the phosphine free
system (Figure SI20h and SI20i in the Supporting Information).
by 108 % after 20 h (Figure 4) and by even 155 % after 48 h of
light irradiation as compared to the phosphine free catalytic
reaction (Table SI4 in the Supporting Information).
Figure 4. Evolved H2 gas after 20 h of light irradiation in the original,
phosphine free system (first data point) and the phosphine containing system
as a function of the applied co-catalyst P(R)3. The detected formation of iron
−
phosphides 1P(R)3 (green) and 2COP(R)2 (violet) is denoted by the color of
the bar in case of their formation during experiments with the respective co-
catalyst. Conditions: 0.5 mM Ir-PS, 0.305 mM 3CO, 20 mL THF/TEA/H2O
(ratio: 4/1/1, pH 12), visible light (380-700 nm, 1.5 W), 25 °C, first data
point: 0 mM P(R)3, other: 0.305 mM P(R)3.
Direct application of preformed, molecularly defined
[NEt4][2COP(c)2] or [NEt4][2COP(d)2] as catalyst results in a gas
evolution curve broadly similar to that of the respective in-situ
system 3CO + P(c)3/P(d)3 (Figure 5, black gas evolution curve
for [NEt4][2COP(d)2]), solid gray gas evolution curve for 3CO +
P(d)3). Especially at extended reaction times, the slopes of the
gas evolution curves are equal, because at this point all iron
material of the in-situ assembly 3CO + P(c)3/P(d)3 has been
Figure 3. Gas evolution curves (grey) and concentration curves of H3− (red),
−
−
converted into 2COP(c)2 or 2COP(d)2 and the solution
composition is similar to that of the molecularly defined system.
However, direct application of the molecular defined catalyst
leads to an initial gas evolution rate increased by 80 %
compared to the in-situ system (Table 3). We therefore conclude
that photons and thus electrons coming from the Ir-PS are more
−
H1− (blue), 1P(d)3 (green) and 2COP(d)2 (violet) in the original, phosphine
free proton reduction system with 3CO as catalyst precursor (top) and the
phosphine containing system with 3CO and the co-catalyst P(d)3 (bottom).
Conditions: 0.5 mM Ir-PS, 0.305 mM 3CO, 20 mL THF/TEA/H2O (ratio: 4/1/1,
pH 12), visible light (380-700 nm, 1.5 W), 25 °C, top (a): 0 mM P(d)3, bottom
(b): 0.305 mM P(d)3.
−
−
efficiently used for proton reduction by 2COP(c)2 or 2COP(d)2
than by the H3−/H1− couple, which is primarily present in the
phosphine free system or in the starting phase of the in-situ
system 3CO + P(c)3/P(d)3. Notably, the higher initial rates in
case of application of the molecularly defined catalysts lead to
H2 yields enhanced by up to 10 % as compared to the in-situ
catalysts. For [NEt4][2COP(c)2], the highest TON[41] was found
with 1111 turnovers after 20 hours at optimized reaction
conditions (for details see Table 3). IR spectra of the reaction
solution with directly applied [NEt4][2COP(d)2] show predominant
In contrast, the application of the electron poor arylphosphines
P(c)3 or P(d)3, highly improves the system due to the formation
−
−
of 2COP(c)2 or 2COP(d)2 . We illustrate this for the example of
P(d)3 (Figure 3, bottom): At the beginning of the experiment, H1−
(Figure 3, bottom blue concentration curve) is the major iron
species in solution. Thus, at this point, the system shows similar
catalytic activity as the phosphine free system. However, during
the first six hours of the reaction, 2COP(d)2− emerges (Figure 3,
bottom violet concentration curve) and progressively contributes
to gas evolution (Figure 3, bottom grey gas evolution curve). The
complex is more stable than H1− and maintains H2 production
−
contributions for 2COP(d)2 throughout the active phase of the
system. Only minor, very weak, IR bands at 2043, 1988, 1956,
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
1890 and 1871 cm−1 of yet unassigned species are observed in
the same time span as well as H1− in very low concentration at
the beginning of the experiment (Figure SI22z and SI23 in the
Supporting Information). On the basis of these experiments,
was found to dissociate from the PS and undergo P-C cleavage.
TONs similar to those of the phosphine free Ir-PS system were
observed.
−
−
2COP(c)2 and 2COP(d)2 are confirmed to be catalytically
highly important key species. Our results suggest that these
complexes can be regarded as a mimic of the [FeFe]-H2ase
active site since they are dinuclear iron complexes that show
catalytic activity and possess carbonyls as well as two bridging
ligands in butterfly conformation.
Table 3. Precursor 3CO, the modified system 3CO + P(R)3 or the preformed
catalyst [NEt4][2COP(R)2] applied in the light driven hydrogen generation at
optimized conditions[a]. Initial Gas Evolution Rate and initial TOF of the catalyst
during the first 0.5 h of light irradiation as well as evolved gas and TON’s of
the catalyst after 20 h light irradiation.
[b]
[b]
Entry Precursor or
Catalyst
Initial
Rate
/ L•h−1
0.5 h
TOFcat
Evolved
Gas
/ mL
TONcat
/ h−1
0.5 h
20 h
20 h
1
3CO
21
21
(260)
33
(408)
2 [c]
3CO + P(c)3
172
130
1073
(260)
(1610)
3
3CO + P(d)3
21
172
120
980
(260)
(1485)
4
5
6
7
[NEt4][2COP(c)2]
[NEt4][2COP(d)2]
[NEt4][2COP(e)2]
[NEt4][2COP(f)2]
37
38
31
27
306
309
257
222
136
134
80
1111
1095
653
49
400
Figure 5. Comparison of the performances of the molecularly defined catalyst
[NEt4][2COP(d)2] (solid black), the in-situ system 3CO + P(d)3 (solid grey) and
3CO (dashed grey) in the light driven hydrogen generation at optimized
conditions: 1.5 mM Ir-PS, 10 mL THF/TEA/H2O (ratio: 3/2/1, pH 12), 20 h Xe-
light irradiation (1.5 W, no filter), 25 °C, solid black: 0.5 mM [NEt4][2COP(d)2],
solid grey: 0.33 mM 3CO + 0.5 mM P(d)3, dashed grey: 0.33 mM 3CO.
[a] conditions: 1.5 mM Ir-PS, 10 mL THF/TEA/H2O (ratio: 3/2/1, pH 12), 20 h
Xe-light irradiation (1.5 W, no filter), 25 °C, entry 1: 0.33 mM 3CO, entry 2-3:
0.33 mM 3CO and 0.5 mM P(R)3, entry 4-7: 0.5 mM [NEt4][2COP(R)2]. [b] TON
= nH2/ncat, TOF = nH2/ncat/t with ncat = amount of 2COP(R)2− or P(R)3, since the
−
quantity of P(R)3 determines the max. possible amount of 2COP(R)2 to be
formed in the in-situ system (in parenthesis: ncat = amount of 3CO) [c] values
taken from ref[14a]: TON after 24 h, TOF during first 3 h.
−
However, the activity of compounds with 2COP(R)2 structure
highly depends on the electron acceptor/donor ability of the
substituent R. This is shown by application of precursor 3CO
together with the co-catalysts P(e)3 or P(f)3. During the reaction,
Deactivation Processes during photocatalytic water
reduction: As described in our previous works, the phosphine
free system is deactivated by decomposition of H1− (Figure 6,
top, blue curve). Decomposition mechanisms were found to be
light induced decarbonylation and poisoning of the catalyst by
bipyridine (bpy). The latter process occurs at high
concentrations of the Ir-PS (Ir/Fe ratio > 0.4), which degrades
over reaction time and releases bpy due to light irradiation. This
was shown by Bernhard and co-workers by means of ESI-MS
spectrometry.[44] The bpy ligand is transferred to the catalyst
resulting in the formation of less active [Fe(CO)3(bpy)] (Figure 6,
top, orange curve). The assignment of this species was
discussed in our previous work and is based on the good
agreement between the wavenumbers of our in-situ experiments
and those quoted in the literature and obtained from DFT
calculations.[14d]
−
−
electron rich 2COP(e)2 or 2COP(f)2 are formed. In contrast to
−
−
their electron poor counterparts 2COP(c)2 and 2COP(d)2 , they
even cause a decrease of the system’s H2 yield (Figure 4).
Direct
application of
preformed [NEt4][2COP(e)2]
or
[NEt4][2COP(f)2] accordingly results in lower TONs as compared
to [NEt4][2COP(c)2] or [NEt4][2COP(d)2] (Table 3).
−
Hence, we conclude the formation of 2COP(R)2 to be
necessary but not sufficient to improve the performance of
precursor 3CO by addition of P(R)3. For the latter, R has to meet
a further requirement: (iii) R must be electron withdrawing.
This activity dependence on R is consistent with findings
described in literature. S.P. Best and co-workers found
−
2COP(e)2 to be an inactive side product during the
electrochemical proton reduction by [Fe2(CO)6(-P(Ph)2)2].[24b, 42]
−
In our previous works, the formation of 2COP(e)2 was also
−
In contrast to H1−, catalyst 2COP(R)2 is inert to bpy poisoning
observed in photocatalytic proton reduction by a non-noble-
metal based system introduced by Beller and co-workers.[43] The
since formation of [Fe(CO)3(bpy)] is not observed in experiments
at high Ir-PS concentration with [NEt4][2COP(d)2] as preformed
catalyst (Figure 6, bottom, solid curves). This property is a major
reason for the high stability of the phosphine containing system
system
uses
a
copper
complex
as
PS
([Cu(xantphos)(bathocuproine)]PF6) and 3CO as catalyst. The
phosphorus ligand in 2COP(e)2− originates from xantphos, which
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
−
compared to the phosphine free one. Nevertheless, 2COP(R)2
Deactivation by excess of phosphine: Beller and co-workers
observed a decline in activity if the catalytic reaction is
performed with more than one equivalent of electron poor
phosphine respective to 3CO. We elucidated this issue and
studied the system applying 3CO and three equivalents of P(d)3.
At these conditions, we found that in a portion of the emerging
is exposed to other slow light induced decomposition processes
such as CO dissociation[45]: While being stable in the dark,
[NEt4][2COP(d)2] disintegrates upon light irradiation in pure THF
in the absence of the PS (Figure 6, bottom, dotted curve).
Remarkably, the process is attenuated in the THF/TEA/H2O
mixture (Figure 6, bottom, dashed curve).[46] Under catalytic
conditions with Ir-PS present, the decomposition further slows
down at extended reaction times (Figure 6, bottom, solid violet
curve).[47] Parallel to catalyst decomposition, the concentration of
the Ir-PS decays due to light irradiation.[14a, 44] That is why the H2
evolution rate steadily declines throughout the reaction while
−
2COP(d)2 complex a terminal carbonyl group was replaced by
one P(d)3 ligand (Scheme 4). As a consequence, the compound
−
[Fe2(CO)5(P(d)3)(-CO)(-P(d)2)]− (2P(d)3P(d)2 ) is formed and
shows IR bands at 1988 (m), 1932 (s), 1911 (m), 1896 (m) and
1872 cm−1 (sh) (Figure 1,left, c, in-situ spectrum at t = 20 h;
Figure 1, middle, turquoise pure component spectrum, Figure 1,
right, turquoise labeled structure). We could assign this species,
−
2COP(R)2 is still present. Accordingly, the activity can be
restored by addition of a second Ir-PS batch at extended
reaction times (Figure SI24 in the Supporting Information).
based on the agreement of its IR spectrum with those of similar
[48]
complexes like [Fe2(CO)5(PMePh2)(-CO)(-PPh2)]−
or
[24b, 49]
[Fe2(CO)5(PHPh2)(-CO)(-PPh2)]−
(Table SI3 in the
Supporting Information). DFT calculation of distinct isomers of
2P(d)3P(d)2− indicates that the position of the terminal phosphine
ligand is in trans position to the phosphido bridge (Figure SI13 in
the Supporting Information).
−
Scheme 4. Substitution of a terminal CO ligand in 2COP(d)2 by P(d)3 yields
the complex 2P(d)3P(d)2−. This reaction is induced by light irradiation.
Figure 6. Top: Performance (grey gas evolution curve) of preformed
[NEt4][H1] (blue concentration curve) applied at high Ir-PS concentration.
Deactivation occurs due to catalyst transformation into Fe(CO)3bpy (orange
concentration curve and structure). Bottom: Performance (grey gas evolution
curve) and concentration curve of molecularly defined [NEt4][2COP(d)2] (violet)
applied at high Ir-PS concentration in THF/TEA/H2O (solid) and concentration
curve of [NEt4][2COP(d)2] applied without Ir-PS in THF/TEA/H2O (dashed) or
in THF only (dotted). Conditions: 0.4 mM [NEt4][H1] (top) or 0.2 mM
[NEt4][2COP(d)2] (bottom), 1.25 mM Ir-PS (solid) or 0 mM Ir-PS (dashed,
dotted), 20 mL THF/TEA/H2O (ratio: 4/1/1, pH 12, solid and dashed) or 10 mL
THF (dotted), visible light (380-700 nm, 1.5 W), 25 °C.
−
Figure 7. Gas evolution curves (grey) and absorption curves of 2COP(d)2
−
(violet) and 2P(d)3P(d)2 (cyan) in the phosphine containing system starting
with precursor 3CO and 1 equivalent of P(d)3 (solid) or 3 equivalents of P(d)3
(dotted). Conditions: 0.5 mM Ir-PS, 0.305 mM 3CO, 0.305 mM P(d)3 (solid) or
0.915 mM P(d)3 (dotted), 20 mL THF/TEA/H2O (ratio: 4/1/1, pH 12), visible
light (380-700 nm, 1.5 W), 25 °C.
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
−
We also generated a mixture of 2P(d)3P(d)2− and 2COP(d)2 , by
another reaction route namely by irradiating a solution of
[NEt4][2COP(d)2] and P(d)3 in pure THF with UV-light (320 - 400
nm) for 30 minutes (Figure SI5 in the Supporting Information).
The NMR spectrum of this solution shows two doublet signals
for 2P(d)3P(d)2 at  131.6 (for -Pd2) and 81.7 ppm (for t-Pd3)
We were also interested in the performance of [NEt4][2COP(d)2]
in electrocatalytic proton reduction. For similar complexes like
[Fe2(CO)6(-P(Ph)2)2] and [Fe2(CO)6(-PPh(CH2)3PPh)], Best
and co-workers observed catalytic currents at -2.25 V vs. Fc0/+
42]
upon addition of tosylic acid (pTsOH).[24b,
In order to get
comparable results we applied the same proton source for the
experiments with [NEt4][2COP(d)2], although recent research
revealed that pTsOH has drawbacks in the sense that it starts to
be directly reduced on glassy carbon working electrodes without
catalyst at −1.5 V.[51]
2
with a coupling constant of JP-P = 36.6 Hz (Figure SI6 in the
Supporting Information), which is consistent with the proposed
structure and the NMR spectra of the aforementioned similar
complexes.
While formation of 2P(d)3P(d)2− occurs only to a minor extend at
one applied equivalent of P(d)3 respective to 3CO under
catalytic conditions (Figure 7, cyan solid absorption curve for
−
2P(d)3P(d)2 ), it plays a significant role at three equivalents of
P(d)3 and leads to 2P(d)3P(d)2− being the major species (Figure
−
7, cyan dotted absorption curve for 2P(d)3P(d)2 ). The additional
phosphine ligand in this complex increases the electron density
−
on the iron, which is indicated by the IR bands of 2P(d)3P(d)2
being found at lower wavenumbers if compared to all 2COP(R)2
−
species.[50] Hence, the electron density on iron in 2P(d)3P(d)2− is
−
even higher than that in the electron rich complexes 2COP(e)2
−
and 2COP(f)2 . As a consequence, H2 evolution is reduced in
the presence of significant amounts of 2P(d)3P(d)2 (Figure 7,
−
grey gas evolution curves). Thus, in order to achieve the highest
possible H2 yield, 3CO and P(R)3 need to be applied in equal
quantities.
Figure 8. Cyclic voltammograms of [NEt4][2COP(d)2] without (violet) and with
additional amounts of tosylic acid (pTsOH). To provide an overview, the
forward scan is denoted as solid line and the backward scan as dotted line.
Electrochemistry: To have an idea of the reasons for the
differing performance of [NEt4][2COP(R)2] in respect of the rest
R, we investigated their electrochemical properties in a brief
preliminary study. The violet graph in Figure 8 shows the cyclic
voltammogram of [NEt4][2COP(d)2] in THF containing
[NBu4][ClO4]. It displays the electrochemical stability of the iron
compound in a potential range between −2.3 and −0.8 V vs.
Fc0/+ (Fc0/+ = ferrocene/ferrocenium redox couple). At more
Parameters: 1 mM [NEt4][2COP(d)2],
0
–
12 mM pTsOH, 0.1 M
(TBA)ClO4/THF,  = 100 mV·s−1, 2 mm2 glassy carbon working electrode, Pt
counter electrode, Ag/AgNO3 reference electrode. Potentials are reported with
respect to the ferrocene/ferrocenium redox couple.
The addition of pTsOH changes the appearance of the cyclic
voltammogram of [NEt4][2COP(d)2]. At 3 equivalents of pTsOH,
the first reduction signal is shifted by 0.24 V to more positive
positive potentials, a quasi-reversible oxidation occurs (E0
=
redox1
−0.58 V vs. Fc0/+). At lower potentials, two reduction steps are
found (peak potentials Ered2 = −2.60 and Ered3 = −2.98 V vs. Fc0/+).
Same signals are found for the other [NEt4][2COP(R)2]
complexes shifted to lower potentials the more electron rich R
gets (Table SI5 in the Supporting Information). A lower reduction
potential corresponds to a lower electron acceptor capability,
which impedes the electron transfer from the Ir-PS to the
catalyst. This might be one possible explanation for the worse
performance of the electron rich [NEt4][2COP(e)2] and
[NEt4][2COP(f)2] in photocatalysis.
values (−2.36
V
vs. Fc0/+). This indicates protonation of
2COP(d)2− prior to reduction, which is confirmed by NMR and IR
spectroscopy. Namely, spectra of a solution of [NEt4][2COP(d)2]
and pTsOH in pure THF show several signals and bands of yet
unassigned species (Figure SI7 and SI8 in the Supporting
Information). However, the high number of these spectral
−
contributions provides indication of a side reaction of 2COP(d)2
with the conjugated base of pTsOH.[52] With increasing
concentration of the proton donor, the first reduction peak is
again shifted catodically. In parallel, the magnitude of the peak
current rises. The latter can be attributed to both the
electrocatalytic proton reduction by the catalyst and the direct
reduction of pTsOH at the electrode overlapping in this potential
range (Figure SI29 in the Supporting Information). However, the
We conceive the reductions steps to be irreversible, since we
did not observe corresponding reoxidation signals with the same
−
magnitude of current density. 2COP(R)2
is perhaps
decomposed at these low potentials. Evidence for the
degradation is indicated by an anodic signal that emerges during
the back scan to positive potentials at −1.72 V in each of the
cyclic voltammograms of the 2COP(R)2− complexes (Figure SI26
and SI27 in the Supporting Information). The signal is also found
at the same position in a cyclic voltammetric measurement of
[Fe(CO)4]2−. The latter would be a potential decomposition
product deriving from degradation of 2COP(R)2− upon reduction.
−
performance of 2COP(d)2 in electrocatalytic proton reduction is
rather low, therefore the catalytic current increase should be
higher. Also, the catalytic peak potential is expected to remain
steady and should be closer to the thermodynamic potential for
proton reduction at −0.95 V vs Fc0/+ (estimated in THF, −0.65 vs
fc0/+ in acetonitrile)[51, 53] offset by only a small overpotential.
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
Hence, 2COP(d)2− in combination with pTsOH does not show an
electrocatalytic performance as good as [Fe2(CO)6(-P(Ph)2)2]
and [Fe2(CO)6(-PPh(CH2)3PPh)].
In contrast to neutral diphosphido bridged diiron compounds like
[Fe2(CO)6(-P(Ph)2)2] investigated by Best and co-workers[24b, 42]
,
[Fe2(CO)6(-CO)(-P(C6H4-4-CF3)2)]− does not show a good
performance in electrocatalytic proton reduction in acidic
electrolyte with tosylic acid as proton source. We attribute the
differing performance of the electrochemical and photochemical
catalysis to the distinct solvent environment, since the
photocatalytic system works in a mixture of THF, triethylamine
and water at pH 12.
The electro-catalytic experiment was performed in acidic
electrolyte whereas photocatalytic water reduction was carried
out in basic solution. Hence, we suppose the differing catalytic
−
activity of 2COP(d)2 to be attributed to the distinct solvent
environment both measurements were performed in.
Heterogeneous nickel alloys, for example, are known to catalyse
the electrocatalytic proton reduction only in an alkaline
electrolyte.[54] Also irradiation and photo excitation of the
protonated catalyst possibly improves the performance in
electrocatalytic proton reduction.[55] We like to point out, that the
processes observed in the electrochemistry experiments can
only be considered as few of many processes occurring during
photocatalysis. Further electrochemical studies with different
reaction conditions have to be carried out in future works.
There are several further possibilities for future improvement or
application of the photocatalytic system: E.g. the use of quantum
dots as photosensitizer or application of water soluble
phosphines like P(C6H4SO3Na)3 (TPPTS), which allows to
increase the water content in the solvent mixture. Immobilization
of the catalyst at polymer supports or electrode surfaces would
allow for its integration into dye-sensitized photoelectrochemical
cells[56]. Such devices dispense with the use of sacrificial
reagents. The immobilization could be realized in-situ e.g. by
application of diarylphosphine-functionalized polysterenes
instead of molecular P(R)3.[28a] Besides, synthesis of derivates
that have not yet been accessible by the self assembly reaction
could be applied as catalyst in molecular systems e.g.
Conclusion
With this detailed spectroscopic study we determined the role of
phosphines which are applied as co-catalyst together with
[Fe3(CO)12] in a photocatalytic proton reduction system of Beller
and co-workers. By means of IR- NMR- and X-ray absorption
spectroscopy, we detected [Fe2(CO)6(-CO)(-P(R)2)]− as key
species in this system. In order that this complex can be formed
in-situ and show a good catalytic activity, the phosphine co-
catalyst has to meet the following requirements:
complexes
of
type
[Fe2(CO)6(-CO)(-P(O^O))]−
or
[Fe2(CO)6(-CO)(-P(R)2)MYx] [24a, 57]. As regards the latter, with
MYx = Cu(bathocuproin) even light harvesting dyads are
conceivable. These ideas illustrate the potential of
Fe2(CO)6(-CO)(-P(R)2)]− as catalyst and its relevance in future
applications.
(i)
(ii)
it belongs to the class of arylphosphines
the ortho positions of the aryl groups are
unsubstituted
Experimental Section
General: All experiments and synthesis procedures were carried out in
an argon atmosphere and under exclusion of air using Schlenk
techniques.
(iii)
the aryl groups are electron poor, e.g. due to electron
withdrawing substituents in meta or para position
Furthermore [Fe3(CO)12] and P(R)3 have to be applied in equal
amounts
since
excessive
P(R)3
deactivates
Chemicals: Solvents were purified and degassed with standard
procedures prior to use. 3CO, P(a-i)3, [NBu4][PF6] and [NBu4][ClO4] were
purchased from a commercial supplier (Sigma-Aldrich, Alfa Aesar, Acros).
[Ir(bpy)(ppy)2]PF6,[58] [HNEt3][HFe3(CO)11][59] and [NEt4[HFe(CO)4][60]
were synthesized according to literature procedures.
[Fe2(CO)6(-CO)(-P(R)2)]− by substitution of a terminal CO
group.
Relative to the original phosphine free system with [HFe3(CO)11]−
and [HFe(CO)4]− as catalytically relevant species, the main
advantages of electron poor [Fe2(CO)6(-CO)(-P(R)2)]− are its
higher durability and resistance against poisoning by
decomposition products of the iridium photosensitizer.
Consequently, the stability of the iridium photosensitizer limits
the yield of the system. The exact mechanism and catalytic
cycle of the proton reduction by [Fe2(CO)6(-CO)(-P(R)2)]− is
not known yet and has to be elucidated in future works.
We classify [Fe2(CO)6(-CO)(-P(R)2)]− as sulfur free [FeFe]-
H2ase mimic. Due to its self assembly (even in the absence of
the photosensitizer), time consuming synthesis is not necessary.
This allows for easy regeneration of the catalyst being important
for the longevity of modern solar fuel devices.[1b]
[NEt4][2COP(e)2] was synthesized by a procedure reported by Walter et
al.[24a]
Synthesis
of
the
other
diferrates
[NEt4][2COP(c)2],
[NEt4][2COP(d)2] and [NEt4][2COP(f)2] started out from preparation of
[NEt4]2[Fe2(CO)8][61]. The subsequent reaction with PClR2 was carried out
according to a procedure similarly described by Osterloh and Reina et
al.:[25, 62] 0.6 mmol of [NEt4]2[Fe2(CO)8] were dispersed in 18 mL THF at
50 °C. After addition of 0.6 mmol of PClR2, the deep red slurry was
stirred for 30 min. Subsequently, further 0.15 mmol PClR2 were added
and stirring was maintained for another 10 min. The solution was then
irradiated by UV light (320 – 500 nm) for 2 hours at room temperature.
During this time, the colour turned from deep red to deep brown. After
filtration, the solution was concentrated to dryness and the solid was
suspended in 5 mL pentane, filtered and washed with additional 5 mL
pentane. The resulting brown residue was dissolved in a mixture of 1.3
mL CH2Cl2 and 1 mL diethyl ether. Pentane was added till the product
precipitated as brownish yellow crystals. The latter were filtered of,
washed with pentane, and dried under vacuum. Yield: 35 – 67 %.
The system is improved by direct application of molecularly
defined [NEt4][Fe2(CO)6(-CO)(-P(R)2)] instead of the in-situ
system increasing the rate and yield of the proton reduction with
a highest initial TOFcat of 306 h−1 and TONcat 20 h of 1111 for R =
C6H4-3,5-(CF3)2.
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
[NEt4][2COP(c)2]: 1H NMR (300 MHz, THF-d8):  = 8.15 (d, 4H, J = 10.3
Hz), 7.93 (s, 2H), 3.23 (m, 8H), 1.21 ppm (m, 12H). 19F NMR (282 MHz,
THF-d8):  = −63.2 ppm (s). 31P NMR (121.5 MHz, THF-d8):  = 122.3
ppm (s). IR (THF): ṽ = 2025 (m), 1977 (s), 1947 (m), 1928 (m), 1747 cm−1
(w).
that of H1−. In Figure 7, absorption curves indicate the maximum of each
species’ absorption spectrum as a function of time.
Hydrogen evolution experiments at optimized conditions (Figure 5, Table
3) were performed in a similar experimental setup that has already been
described elsewhere by the Beller group[13-14] It differs from the setup
described in the previous paragraphs by absence of the microannular
gear pump and the IR unit. A 300 W Xe lamp (Lot-Oriel-300-W-Xe-lamp,
LSB530) was used as light source. At the end of these experiments, the
collected gas was analyzed by GC (gas chromatograph HP6890N,
carboxen 1000, TCD or TCD + methanizer/FID, external calibration).
Apart from argon, traces of solvent vapour and CO dissociated from the
WRC, only hydrogen was found in the gas mixture. Turnover numbers
have been calculated according to TON = VH2/Vm/ncatalyst with
Vm = 24.465 l•mol−1 (molar volume of an ideal gas at 1 atmosphere and
298.15 K).
[NEt4][2COP(d)2]: 1H NMR (300 MHz, THF-d8):  = 7.79 (t, 4H, J = 9.0
Hz), 7.53 (d, 4H, J = 7.7 Hz), 3.27 (q, 8H, J = 7.2 Hz), 1.27 ppm (t, 12H, J
= 7.2 Hz). 19F NMR (282 MHz, THF-d8):  = −62.8 ppm (s). 31P NMR
(121.5 MHz, THF-d8):  = 122.8 ppm (s). IR (THF): ṽ = 2021 (m), 1971 (s),
1942 (m), 1922 (m), 1741 cm−1 (w).
[NEt4][2COP(e)2]: 1H NMR (300 MHz, THF-d8):  = 7.61 (m, 4H), 7.17 (m,
6H), 3.26 (q, 8H, J = 7.3 Hz), 1.26 ppm (t, 12H, J = 7.3 Hz). 31P NMR
(121.5 MHz, THF-d8):  = 125.9 ppm (s). IR (THF): ṽ = 2016 (m), 1965 (s),
1934 (m), 1916 (m), 1736 cm−1 (w).
[NEt4][2COP(f)2]: 1H NMR (300 MHz, THF-d8):  = 7.50 (dd, 4H, J = 10.4
and 8.8 Hz), 6.73 (dd, 4H, J = 8.8 and 1.5 Hz), 3.72 (s, 6H), 2.79 (m, 8H),
0.82 ppm (m, 12H). 31P NMR (121.5 MHz, THF-d8):  = 123.2 ppm (s). IR
(THF): ṽ = 2014 (m), 1963 (s), 1931 (m), 1913 (m), 1736 cm−1 (w).
NMR spectra were recorded with a Bruker Avance AV-300 (1H: 300 MHz,
31P: 121 MHz) or a Bruker Avance AV-400 (1H: 400 MHz, 31P: 162 MHz)
instrument, respectively. 31P-NMR shifts are proton-decoupled and
-
Experimental Setup, Procedure and Data Analysis: The experimental
setup (Scheme SI1 in the Supporting Information) for time-resolved
operando continuous flow FTIR experiments under light irradiation
consists of a reaction vessel, connected to a FTIR-spectrometer and to
an automatic gas burette. Hence, this setup allows for tracking of gas
evolution and the simultaneous acquisition of IR-spectra. A mercury
vapour lamp (LUMATEC) is attached to the double-walled
thermostatically controlled reaction vessel at a fixed position via a fibre
optic cable. This lamp emits visible light between 380 and 700 nm. The
reaction solution is stirred by a magnetic stirrer at constant and
reproducible speed and is continuously circulated by a microannular gear
pump (HNP) through the thermostated measuring cell of the IR-
spectrometer. The cell is equipped with CaF2 windows and has an optical
path length of 100 µm. The FTIR-measurements are carried out on a
Bruker Tensor 27 spectrometer with a mercury-cadmium-telluride (MCT)
detector. Tracking of gas evolution is realized by an automatic gas
burette (MESSEN NORD) at a constant pressure of 1020 mbar. Its
temperature is kept constant at 25 °C. A condenser is installed in
between the reaction vessel and the gas burette and runs at 5 °C to
retain volatile solvent.[63]
reported in parts per million downfield from H3PO4 . 19F-NMR spectra are
measured with the Ir-PS as internal standard (signals at  −71.80 and
−74.31 ppm). The chemical shifts are reported in ppm downfield from
CFCl3.
The X-ray absorption measurements were carried out at the synchrotron
ANKA. The synchrotron beam current was between 80–140 mA at 2.5
GeV storage ring energy. A Si(111) double crystal monochromator was
used for measurements at the Fe K-edge (7.112 keV). The second
monochromator crystal was tilt for optimal harmonic rejection. To perform
operando studies the spectra were recorded in fluorescence mode using
a hyperpure germanium detector. Energy calibration was performed with
an iron metal foil prior to the measurements. Samples in solution were
measured in a specially designed fluorescence cell, which allowed
securation and filling under inert gas flow.[65] An argon atmosphere was
applied to the cell in course of the measurements in order to allow
evolving gas to escape the cell. Details of the analysis are described
elsewhere.[66] EXAFS data analysis was performed according to the
curved wave formalism of the EXCURV98 program with XALPHA phase
and amplitude functions.[67] The amplitude reduction factor (AFAC) was
allowed to float in the fit. Fourier filtered, backtransformed spectra were
analyzed in k-space. Fourier filtering was applied in the range between
1.0 – 3.2 Å.
A typical experiment is carried out as follows: The whole apparatus is
evacuated and purged with argon five times to provide oxygen-free
conditions. The reaction vessel is filled with 20 mL of solvent. After
temperature adjustment an IR-spectrum is taken and used as
background. The solid catalysts are added to the system. After reaching
thermal equilibrium between gas- and condensed phase (10 min), the
reaction is started irradiating the solution (1500 mW). An IR-spectrum
with 64 scans is taken every 2 minutes. The monitored hydrogen
evolution curve is corrected by the system´s blind value, which was
obtained in an experiment without catalysts. Turnover numbers have
Cyclic voltammetry experiments were carried out with a Metrohm Autolab
PGSTAT302N or PalmSens EmStat3+ Blue potentiostat. Electrodes of
the company ALS were used: 2 mm2 glassy carbon disc working
electrode, Pt wire counter electrode, Ag/AgNO3 reference electrode
(silver wire in 0.01 M AgNO3/ 0.1 TBAClO4/ acetonitrile). The reference
electrode was separated from the experimental solution with a vycor frit.
Ferrocene as internal standard was added at the end of each experiment.
been
calculated
according
to
TON = VH2/Vm/ncatalyst
with
Vm = 24.3035 l•mol−1 (molar volume of an ideal gas at 1020 mbar and
298.15 K).
Acknowledgements
Extraction of the IR concentration profiles and determination of the IR
extinction coefficients were carried out by alignment of pure component
spectra onto the in-situ IR spectra. The pure component spectra were
obtained from solution of synthesized compounds (H3−, H1− and
2COP(c-f)2−) or by substraction of these from recorded in-situ spectra
(1P(R)3 and 2P(d)3P(d)2−). Spectrum and concentration curve of
Fe(CO)3(bpy) was extracted by an algorithm based on factor analysis.[64]
The extinction coefficient of Fe(CO)3(bpy) was presumed to be equal to
This work has been partly supported by the BMBF within the
project “Light2Hydrogen” (Spitzenforschung und Innovation in
den Neuen Ländern), by the European Union (European Social
Funds, ESF) within the project “PS4H” and by the Ministry for
Education, Science and Culture of Mecklenburg-Vorpommern.
We thank Wolfgang Baumann for supporting the NMR
measurements, Robert Francke, Felix Gärtner, Michael Karnahl,
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
Multiplying
this
factor
with
the
TON
for
[Fe2(-
Nils Rockstroh, and Petra Bartels for providing chemicals and
expertise, Peter Kumm and Martin Riedel for technical support,
as well as Anika Wilhelms and Elisabeth Oberem for
experimental support. Matthias Bauer thanks the BMBF for
support within the project “SusChEmX” and “TrExHigh” and the
synchrotron ANKA for provision of beamtime.
S(CH2)2SO3Na)2(CO)6] still results in a TON being lower than those
observed in the phosphine containing system from Beller and co
workers.
[13]
[14]
F. Gärtner, B. Sundararaju, A.-E. Surkus, A. Boddien, B. Loges, H.
Junge, P. H. Dixneuf, M. Beller, Angewandte Chemie International
Edition 2009, 48, 9962-9965.
a) F. Gärtner, A. Boddien, E. Barsch, K. Fumino, S. Losse, H.
Junge, D. Hollmann, A. Brückner, R. Ludwig, M. Beller, Chemistry
– A European Journal 2011, 17, 6425-6436; b) D. Hollmann, F.
Gärtner, R. Ludwig, E. Barsch, H. Junge, M. Blug, S. Hoch, M.
Beller, A. Brückner, Angewandte Chemie International Edition
2011, 50, 10246-10250; c) A. Neubauer, G. Grell, A. Friedrich, S. I.
Bokarev, P. Schwarzbach, F. Gärtner, A.-E. Surkus, H. Junge, M.
Beller, O. Kühn, S. Lochbrunner, The Journal of Physical
Chemistry Letters 2014, 5, 1355-1360; d) S. Fischer, O. S.
Bokareva, E. Barsch, S. I. Bokarev, O. Kühn, R. Ludwig,
ChemCatChem 2016, 8, 404-411.
Keywords: Hydrogenase • Photocatalysis • Proton Reduction •
time-resolved IR spectroscopy • EXAFS
[1]
a) N. Armaroli, V. Balzani, Angewandte Chemie International
Edition 2007, 46, 52-66; b) L. Hammarström, Chem 2016, 1, 515-
518; c) N. Armaroli, V. Balzani, Chemistry – A European Journal
2016, 22, 32-57.
[2]
a) R. Eisenberg, Science 2009, 324, 44-45; b) H. Junge, N.
Rockstroh, S. Fischer, A. Brückner, R. Ludwig, S. Lochbrunner, O.
Kühn, M. Beller, Inorganics 2017, 5, 14.
[15]
[16]
F. Gärtner, D. Cozzula, S. Losse, A. Boddien, G. Anilkumar, H.
Junge, T. Schulz, N. Marquet, A. Spannenberg, S. Gladiali, M.
Beller, Chemistry – A European Journal 2011, 17, 6998-7006.
J. A. S. Howell, N. Fey, J. D. Lovatt, P. C. Yates, P. McArdle, D.
Cunningham, E. Sadeh, H. E. Gottlieb, Z. Goldschmidt, M. B.
Hursthouse, M. E. Light, Journal of the Chemical Society, Dalton
Transactions 1999, 3015-3028.
[3]
[4]
Z. Han, R. Eisenberg, Accounts of Chemical Research 2014, 47,
2537-2544.
a) P. Du, R. Eisenberg, Energy & Environmental Science 2012, 5,
6012-6021; b) A. C. Brooks, K. Basore, S. Bernhard, Chemical
Communications 2014, 50, 5196-5199; c) H. Junge, Z. Codolà, A.
Kammer, N. Rockstroh, M. Karnahl, S.-P. Luo, M.-M. Pohl, J.
Radnik, S. Gatla, S. Wohlrab, J. Lloret, M. Costas, M. Beller,
Journal of Molecular Catalysis A: Chemical 2014, 395, 449-456; d)
W. T. Eckenhoff, Coordination Chemistry Reviews 2017.
a) J. W. Peters, W. N. Lanzilotta, B. J. Lemon, L. C. Seefeldt,
Science 1998, 282, 1853-1858; b) Y. Nicolet, C. Piras, P. Legrand,
C. E. Hatchikian, J. C. Fontecilla-Camps, Structure 1999, 7, 13-23;
c) Y. Nicolet, A. L. de Lacey, X. Vernède, V. M. Fernandez, E. C.
Hatchikian, J. C. Fontecilla-Camps, Journal of the American
Chemical Society 2001, 123, 1596-1601; d) A. Silakov, B. Wenk, E.
Reijerse, W. Lubitz, Physical Chemistry Chemical Physics 2009,
11, 6592-6599; e) G. Berggren, A. Adamska, C. Lambertz, T. R.
Simmons, J. Esselborn, M. Atta, S. Gambarelli, J. M. Mouesca, E.
Reijerse, W. Lubitz, T. Happe, V. Artero, M. Fontecave, Nature
2013, 499, 66.
[17]
[18]
In a previous article [ref. 14a.], a sample of 3CO + P(c)3 dissolved
in THF/TEA/H2O was also measured by 31P-NMR spectroscopy.
The spectrum shows a signal at  84.7 ppm which most probably
belongs to 1P(c)3 rather than [Fe3(CO)12P(c)3] as assumed in the
previous article (see Supporting Information, Table SI1).
a) A. F. Clifford, A. K. Mukherjee, Inorganic Chemistry 1963, 2,
151-153; b) S. M. Grant, A. R. Manning, Inorganica Chimica Acta
1978, 31, 41-48; c) A. Shojaie, J. D. Atwood, Organometallics
1985, 4, 187-190.
[5]
[19]
[20]
S. B. Butts, D. F. Shriver, Journal of Organometallic Chemistry
1979, 169, 191-197.
In the operando IR experiments of the work, an optical filter is
installed in front of the mercury light source and the reaction
solution is irradiated with visible light having wavelengths in the
range of 380 – 700 nm.
[6]
a) J.-F. Capon, F. Gloaguen, P. Schollhammer, J. Talarmin,
Coordination Chemistry Reviews 2005, 249, 1664-1676; b) S.
Tschierlei, S. Ott, R. Lomoth, Energy & Environmental Science
2011, 4, 2340-2352; c) N. Wang, M. Wang, L. Chen, L. Sun,
Dalton Transactions 2013, 42, 12059-12071; d) W. Lubitz, H.
Ogata, O. Rüdiger, E. Reijerse, Chemical Reviews 2014, 114,
4081-4148; e) D. Schilter, J. M. Camara, M. T. Huynh, S.
Hammes-Schiffer, T. B. Rauchfuss, Chemical Reviews 2016, 116,
8693-8749; f) M. Watanabe, Y. Honda, H. Hagiwara, T. Ishihara,
Journal of Photochemistry and Photobiology C: Photochemistry
Reviews 2017, 33, 1-26.
[21]
[22]
a) G. Bor, Inorganica Chimica Acta 1967, 1, 81-92; b) M. Y.
Darensbourg, H. L. Conder, D. J. Darensbourg, C. Hasday,
Journal of the American Chemical Society 1973, 95, 5919-5924; c)
D. J. Darensbourg, H. H. Nelson, C. L. Hyde, Inorganic Chemistry
1974, 13, 2135-2145.
a) H. L. Conder, M. Y. Darensbourg, Inorganic Chemistry 1974, 13,
506-511; b) M. O. Albers, N. J. Coville, T. V. Ashworth, E.
Singleton, Journal of Organometallic Chemistry 1981, 217, 385-
390; c) L. R. Martin, F. W. B. Einstein, R. K. Pomeroy, Inorganic
Chemistry 1985, 24, 2777-2785; d) J. E. Ellis, Y. S. Chen,
Organometallics 1989, 8, 1350-1361; e) J. A. S. Howell, J. D.
Lovatt, P. McArdle, D. Cunningham, E. Maimone, H. E. Gottlieb, Z.
Goldschmidt, Inorganic Chemistry Communications 1998, 1, 118-
120; f) N. J. Farrer, R. McDonald, J. S. McIndoe, Dalton
Transactions 2006, 4570-4579.
[7]
[8]
[9]
P. Zhang, M. Wang, Y. Na, X. Li, Y. Jiang, L. Sun, Dalton
Transactions 2010, 39, 1204-1206.
T. Yu, Y. Zeng, J. Chen, Y.-Y. Li, G. Yang, Y. Li, Angewandte
Chemie International Edition 2013, 52, 5631-5635.
J.-X. Jian, Q. Liu, Z.-J. Li, F. Wang, X.-B. Li, C.-B. Li, B. Liu, Q.-Y.
Meng, B. Chen, K. Feng, C.-H. Tung, L.-Z. Wu, Nature
Communications 2013, 4, 2695.
[23]
[24]
C. A. Tolman, Chemical Reviews 1977, 77, 313-348.
a) B. Walther, H. Hartung, H.-C. Böttcher, U. Baumeister, U.
Böhland, J. Reinhold, J. Sieler, J. Ladriere, H.-M. Schiebel,
Polyhedron 1991, 10, 2423-2435; b) M. H. Cheah, S. J. Borg, M. I.
Bondin, S. P. Best, Inorganic Chemistry 2004, 43, 5635-5644.
W. T. Osterloh, Ph.D. thesis, The University of Texas (Austin),
1982.
[10]
[11]
[12]
J.-X. Jian, C. Ye, X.-Z. Wang, M. Wen, Z.-J. Li, X.-B. Li, B. Chen,
C.-H. Tung, L.-Z. Wu, Energy & Environmental Science 2016, 9,
2083-2089.
J. I. Goldsmith, W. R. Hudson, M. S. Lowry, T. H. Anderson, S.
Bernhard, Journal of the American Chemical Society 2005, 127,
7502-7510.
[25]
[26]
a) E. P. Kyba, R. E. Davis, C. N. Clubb, S.-T. Liu, H. O. A. Palacios,
J. S. McKennis, Organometallics 1986, 5, 869-877; b) R. T. Baker,
both, Ru-PS and Ir-PS belong to the same class. Nevertheless,
the former has a lower activity than the latter by a factor of 8.
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
J. C. Calabrese, P. J. Krusic, M. J. Therien, W. C. Trogler, Journal
Lochbrunner, R. Ludwig, A. Brückner, ACS Catalysis 2014, 4,
of the American Chemical Society 1988, 110, 8392-8412.
A. Rahaman, F. R. Alam, S. Ghosh, M. Haukka, S. E. Kabir, E.
Nordlander, G. Hogarth, Journal of Organometallic Chemistry 2013,
730, 123-131.
1845-1849; d) A. J. J. Lennox, S. Fischer, M. Jurrat, S.-P. Luo, N.
[27]
[28]
Rockstroh, H. Junge, R. Ludwig, M. Beller, Chemistry
European Journal 2016, 22, 1233-1238.
– A
[44]
L. L. Tinker, N. D. McDaniel, P. N. Curtin, C. K. Smith, M. J. Ireland,
S. Bernhard, Chemistry – A European Journal 2007, 13, 8726-
8732.
a) P. E. Garrou, Chemical Reviews 1985, 85, 171-185; b) P. W. N.
M. van Leeuwen, Applied Catalysis A: General 2001, 212, 61-81;
c) A. W. Parkins, Coordination Chemistry Reviews 2006, 250, 449-
467; d) S. A. Macgregor, Chemical Society Reviews 2007, 36, 67-
76.
[45]
[46]
B. T. Cooper, J. A. Johnson, S. W. Buckner, Inorganic Chemistry
1995, 34, 5375-5377.
In both solvent environments (THF or THF/TEA/H2O) in the
absence of Ir-PS and under light irradiation, weak IR contributions
[29]
[30]
a) L. W. Gosser, Inorganic Chemistry 1977, 16, 430-434; b) M. A.
Alvarez, C. Alvarez, M. E. García, V. Riera, M. A. Ruiz, C. Bois,
Organometallics 1997, 16, 2581-2589.
−
of transient species originating from decomposition of 2COP(d)2
were detected at 2000, 1933, 1920 and 1895 cm−1
.
a) F. H. Westheimer, S. Huang, F. Covitz, Journal of the American
Chemical Society 1988, 110, 181-185; b) S. K. McIntyre, T. M.
Alam, Magnetic Resonance in Chemistry 2007, 45, 1022-1026.
A. G. Abatjoglou, E. Billig, D. R. Bryant, Organometallics 1984, 3,
923-926.
[47]
Under catalytic conditions in THF/TEA/H2O with Ir-PS present and
under light irradiation, weak IR contributions of transient species
−
originating from decomposition of 2COP(d)2 were detected at
[31]
[32]
2043, 1988, 1956, 1890 and 1871 cm−1
.
[48]
[49]
[50]
Y.-F. Yu, J. Gallucci, A. Wojcicki, Journal of the Chemical Society,
Chemical Communications 1984, 653-655.
N. M. Doherty, G. Hogarth, S. A. R. Knox, K. A. Macpherson, F.
Melchior, D. A. V. Morton, A. G. Orpen, Inorganica Chimica Acta
1992, 198–200, 257-270.
Y. F. Yu, J. Gallucci, A. Wojcicki, Journal of the American
Chemical Society 1983, 105, 4826-4828.
[33]
[34]
[35]
K. Kiyotomi, S. Kotaro, T. Shiichiro, Chemistry Letters 1979, 8,
821-822.
a) L. E. Orgel, Inorganic Chemistry 1962, 1, 25-29; b) P. S.
Braterman, Metal carbonyl spectra / P. S. Braterman, Academic
Press, London ; New York, 1975.
W. A. Herrmann, C. Broßmer, K. Öfele, M. Beller, H. Fischer,
Journal of Molecular Catalysis A: Chemical 1995, 103, 133-146.
a) M. Bauer, H. Bertagnolli, in Methods in Physical Chemistry,
Wiley-VCH Verlag GmbH & Co. KGaA, 2012, pp. 231-269; b) M. G.
Pfeffer, B. Schäfer, G. Smolentsev, J. Uhlig, E. Nazarenko, J.
Guthmuller, C. Kuhnt, M. Wächtler, B. Dietzek, V. Sundström, S.
Rau, Angewandte Chemie International Edition 2015, 54, 5044-
5048.
[51]
B. D. McCarthy, D. J. Martin, E. S. Rountree, A. C. Ullman, J. L.
Dempsey, Inorganic Chemistry 2014, 53, 8350-8361.
F. Gloaguen, Inorganic Chemistry 2016, 55, 390-398.
G. A. N. Felton, R. S. Glass, D. L. Lichtenberger, D. H. Evans,
Inorganic Chemistry 2006, 45, 9181-9184.
[52]
[53]
[54]
[55]
[56]
[57]
M. Zeng, Y. Li, Journal of Materials Chemistry A 2015, 3, 14942-
14962.
[36]
a) N. Binsted, S. L. Cook, J. Evans, G. N. Greaves, R. J. Price,
Journal of the American Chemical Society 1987, 109, 3669-3676;
b) N. Binsted, J. Evans, G. N. Greaves, R. J. Price, Chem.
Commun. 1987, 0, 1330-1333; c) T. E. Westre, A. Di Cicco, A.
Filipponi, C. R. Natoli, B. Hedman, E. I. Solomon, K. O. Hodgson,
Journal of the American Chemical Society 1995, 117, 1566-1583.
A particular shell denotes an ensemble of backscattering atoms of
the same type at a particular distance from the Fe atom.
J. Reiko, J. Ralf, P. Irene, F. Robert, J. Haijun, B. Matthias, B.
Matthias, ChemSusChem 2012, 5, 734-739.
L. Bertini, P. Fantucci, L. De Gioia, G. Zampella, Inorganic
Chemistry 2013, 52, 9826-9841.
Z. Yu, F. Li, L. Sun, Energy & Environmental Science 2015, 8,
760-775.
a) M. Ferrer, A. Julià, R. Reina, O. Rossell, M. Seco, D. de
Montauzon, Journal of Organometallic Chemistry 1998, 560, 147-
[37]
[38]
[39]
[40]
153; b) M. Ferrer, O. Rossell, M. Seco, M. Soler, M. Font-Bardıa, X.
́
Solans, D. de Montauzon, Journal of Organometallic Chemistry
2000, 598, 215-221; c) M. Friederici, I. Angurell, M. Seco, O.
Rossell, J. Llorca, Dalton Transactions 2011, 40, 7934-7940.
E. D. Cline, S. E. Adamson, S. Bernhard, Inorganic Chemistry
2008, 47, 10378-10388.
D. Braga, L. Farrugia, F. Grepioni, B. F. G. Johnson, Journal of
Organometallic Chemistry 1994, 464, C39-C41.
[58]
[59]
[60]
[61]
a) M. Bauer, H. Bertagnolli, The Journal of Physical Chemistry B
2007, 111, 13756-13764; b) M. Bauer, C. Gastl, C. Köppl, G.
Kickelbick, H. Bertagnolli, Monatsh. Chem. 2006, 137, 567-581; c)
E. Hunstock, C. Mealli, M. J. Calhorda, J. Reinhold, Inorganic
Chemistry 1999, 38, 5053-5060.
W. McFarlane, G. Wilkinson, W. Hübel, in Inorganic Syntheses,
John Wiley & Sons, Inc., 2007, pp. 181-183.
I. Y. Guzman-Jimenez, J. W. van Hal, K. H. Whitmire,
Organometallics 2003, 22, 1914-1922.
a) J. P. Collman, R. G. Finke, P. L. Matlock, R. Wahren, R. G.
Komoto, J. I. Brauman, Journal of the American Chemical Society
1978, 100, 1119-1140; b) T. M. Bockman, H. C. Cho, J. K. Kochi,
Organometallics 1995, 14, 5221-5231.
[41]
To compare this TON with the previous published TONs for the
precursor system 3CO + P(R)3 by Beller et al., one has to
recalculate the latter, since they were calculated as n(H2)/n(3CO)
rather than n(H2)/n(2COP(R)2−). In the system 3CO + P(R)3 the
trinuclear precursor is converted into the dinuclear species
2COP(R)2−. By this, the amount of catalyst rises and the TON
decreases by a factor of 1.5.
[62]
[63]
a) R. Reina, O. Rossell, M. Seco, Journal of Organometallic
Chemistry 1990, 398, 285-291; b) D. Seyferth, K. S. Brewer, T. G.
Wood, M. Cowie, R. W. Hilts, Organometallics 1992, 11, 2570-
2579.
[42]
[43]
M. H. Cheah, S. J. Borg, S. P. Best, Inorganic Chemistry 2007, 46,
1741-1750.
a) B. Loges, H. Junge, B. Spilker, C. Fischer, M. Beller, Chemie
Ingenieur Technik 2007, 79, 741-753; b) A. Boddien, B. Loges, H.
Junge, M. Beller, ChemSusChem 2008, 1, 751-758.
K. Neymeyr, M. Sawall, D. Hess, Journal of Chemometrics 2010,
24, 67-74.
a) S.-P. Luo, E. Mejía, A. Friedrich, A. Pazidis, H. Junge, A.-E.
Surkus, R. Jackstell, S. Denurra, S. Gladiali, S. Lochbrunner, M.
Beller, Angewandte Chemie International Edition 2013, 52, 419-
423; b) E. Mejía, S.-P. Luo, M. Karnahl, A. Friedrich, S. Tschierlei,
A.-E. Surkus, H. Junge, S. Gladiali, S. Lochbrunner, M. Beller,
Chemistry – A European Journal 2013, 19, 15972-15978; c) S.
Fischer, D. Hollmann, S. Tschierlei, M. Karnahl, N. Rockstroh, E.
Barsch, P. Schwarzbach, S.-P. Luo, H. Junge, M. Beller, S.
[64]
[65]
[66]
M. Bauer, C. Gastl, Physical Chemistry Chemical Physics 2010, 12,
5575-5584.
D. Mellmann, E. Barsch, M. Bauer, K. Grabow, A. Boddien, A.
Kammer, P. Sponholz, U. Bentrup, R. Jackstell, H. Junge, G.
This article is protected by copyright. All rights reserved.

10.1002/chem.201802694
Chemistry - A European Journal
FULL PAPER
Laurenczy, R. Ludwig, M. Beller, Chemistry – A European Journal
2014, 20, 13589-13602.
[67]
N. Binsted, S. S. Hasnain, Journal of Synchrotron Radiation 1996,
3, 185-196.
This article is protected by copyright. All rights reserved.