Photochemical Water-Splitting with Organomanganese Complexes

Article abstract of DOI:10.1021/acs.inorgchem.7b01483

Certain organometallic chromophores with water-derived ligands, such as the known [Mn(CO)₃(μ₃-OH)]₄ (1) tetramer, drew our attention as possible platforms to study water-splitting reactions. Herein, we investigate the UV irradiation of various tricarbonyl organomanganese complexes, including 1, and demonstrate that dihydrogen, CO, and hydrogen peroxide form as products in a photochemical water-splitting decomposition reaction. The organic and manganese-containing side products are also characterized. Labeling studies with ¹⁸O-1 suggest that the source of oxygen atoms in H₂O₂ originates from free water that interacts with 1 after photochemical dissociation of CO (1-CO) constituting the oxidative half-reaction of water splitting mediated by 1. Hydrogen production from 1 is the result of several different processes, one of which involves the protons derived from the hydroxido ligands in 1 constituting the reductive half-reaction of water splitting mediated by 1. Other processes that generate H₂ are also operative and are described. Collectively the results from the photochemical decomposition of 1 provide an opportunity to propose a mechanism, and it is discussed within the context of developing new strategies for water-splitting reactions with organomanganese complexes.

Full text of DOI:10.1021/acs.inorgchem.7b01483

Article
pubs.acs.org/IC
Photochemical Water-Splitting with Organomanganese Complexes
Karthika J. Kadassery, Suman Kr Dey, Anthony F. Cannella, Roshaan Surendhran,
and David C. Lacy*
Department of Chemistry, University at Buffalo, State University of New York, Buffalo, New York 14260, United States
S
* Supporting Information
ABSTRACT: Certain organometallic chromophores with
water-derived ligands, such as the known [Mn(CO)3(μ3−
OH)]4 (1) tetramer, drew our attention as possible platforms
to study water-splitting reactions. Herein, we investigate the
UV irradiation of various tricarbonyl organomanganese
complexes, including 1, and demonstrate that dihydrogen,
CO, and hydrogen peroxide form as products in a photo-
chemical water-splitting decomposition reaction. The organic
and manganese-containing side products are also character-
ized. Labeling studies with 18O-1 suggest that the source of
oxygen atoms in H2O2 originates from free water that interacts with 1 after photochemical dissociation of CO (1-CO)
constituting the oxidative half-reaction of water splitting mediated by 1. Hydrogen production from 1 is the result of several
different processes, one of which involves the protons derived from the hydroxido ligands in 1 constituting the reductive half-
reaction of water splitting mediated by 1. Other processes that generate H2 are also operative and are described. Collectively the
results from the photochemical decomposition of 1 provide an opportunity to propose a mechanism, and it is discussed within
the context of developing new strategies for water-splitting reactions with organomanganese complexes.
Chart 1. Organometallic Complexes for Water-Splitting
INTRODUCTION
■
Organometallic compounds and their reactions have been the
focus of many successful mechanistic investigations leading to
industrially practiced chemistry. It follows that efforts in solar
energy storage by means of water-splitting technology could
benefit from the strategic application of organometallic
catalysts.1,2 In particular, a small class of organometallic
chromophores that contain water-derived ligands mediate
light-driven water-splitting transformations producing both a
reduced and oxidized water-derived product. Milstein and co-
workers discovered one of the first of these water-splitting
systems using light and a ruthenium pincer pyridine complex.3,4
At the same time, Kunkely and Vogler uncovered two isolated
oxidation/reduction pathways to produce O2 and H2 from
water using osmocene and light.5,6 Additionally, Beweries and
Rosenthal and co-workers reported a water-splitting half
reaction7 with titanocene and subsequently improved the
system to a fully realized water-splitting cycle.8 Finally, Fan and
co-workers have demonstrated that cymantrene (CpMn(CO)3)
is capable of photochemically transforming water into H2 and
H2O2.9
These examples constitute a rare class of organometallic
compounds that photochemically split water into H2 and an
oxidized water equivalent (Chart 1). These systems are distinct
from other important catalysts10,11 in that both half reactions
nominally occur at a single site, and sacrificial reagents or
sensitizers are not needed. A thorough knowledge of the
mechanism whereby H2 and H2O2 forms by irradiation of the
said class of compounds is of paramount importance for
developing catalytic water-splitting technologies that utilize
organometallic complexes. Generally, the ligand plays an
important role in the reaction mechanism of organometallic
complexes. For instance, the mixed phosphine/amine pyridine
pincer complex that Milstein used for ruthenium-mediated
water-splitting plays a functional role in transporting redox
equivalence in the form of a transferable hydride. In another
example, Beweries showed that the hydrogen production half-
reaction with titanocene can combine with an oxidative half-
reaction if the Cp ligands are tied together in an ansa-
titanocene complex.
Unfortunately, of these four examples all but osmocene suffer
from the fact that H2O2 (or HO·), rather than O2, is the initial
product of water oxidation. In addition to this, the apparent
lack of catalysis is perhaps the greatest drawback of these
transformations. However, Fan and co-workers demonstrated
catalytic production of H2 from thiols with concomitant
disulfide production using MnI-carbonyl compounds.12,13 If
the mechanisms with water splitting and “thiol splitting” are
similar, they differ primarily in the calamitous effects of
Received: June 10, 2017
© XXXX American Chemical Society
A
DOI: 10.1021/acs.inorgchem.7b01483
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
system and stored over 3 Å molecule sieves inside of a glovebox.
Molecular sieves were activated at 200 °C under vacuum (<100
mTorr) for 2−3 d prior to use. Solvents dried by this method
contained ∼10 ppm water. Dryer solvent was required for certain
experiments and was obtained by passing predried solvent through
freshly activated alumina (<100 mTorr, 300 °C, 18 h); these
experiments are explicitly defined in the Results Section. Deuterated
solvents were degassed by freeze−pump−thaw methods and stored
inside a glovebox over 3 Å sieves. The lithium salts LiCptet and LiCp*
were prepared by addition of 1 equiv of n-BuLi (0.25 M in hexane) to
1 M solution of CptetH (1 g, 0.008 mol) or Cp*H (5 mL, 0.0319 mol,
distilled prior to use) in hexane at −78 °C, stirred overnight, and
subsequently filtered and washed with hexane in a glovebox and dried
(80% yield both cases). The following compounds were prepared
according to literature: 1·(C6H6)2;16 (Tp)Mn(CO)3 (Tp = trispyr-
azolyl borate), CptetMn(CO)3, and Cp*Mn(CO)3 were prepared
using (py)2Mn(CO)3Br as the manganese synthon;17 tris(1-
pyrazolyl)methane (Tpm), tris(3,5-dimethyl-1-pyrazoyl)methane
(Tpm*), [(Tpm)Mn(CO)3]OTf, and [(Tpm*)Mn(CO)3]OTf;18
(bipy)Mn(CO)3Br and (phen)Mn(CO)3Br and (phen)Mn-
(CO)3Br.19,20
Synthesis of [Ph3PO]2·H2O2. Open to air, Ph3P (2 g, 0.0076 mol)
was dissolved in 150 mL of toluene in a Schlenk flask, which was
cooled to 0 °C and blanketed with argon. H2O2 (10 equiv; 30% in
H2O, 9 mL, 0.076 mol H2O2) was added dropwise via syringe with
vigorous stirring. The reaction mixture was stirred overnight under
argon and allowed to slowly warm to room temperature. After 12 h,
the solution was filtered open to the air through a fritted glass funnel
to obtain a white precipitate; the material was subsequently dried
under vacuum (room temperature (RT)) for 18 h and brought into a
glovebox. The ATR-FTIR and 1H NMR spectra of the white
precipitate showed Ph3PO·H2O2·H2O adduct according to literature.21
The Ph3PO·H2O2·H2O adduct was further dried by dissolving in dry
acetonitrile, freezing the solution, and removing solvent at RT in
vacuo. Anhydrous [Ph3PO]2·H2O2 was thus obtained as a white
crystalline solid and showed no trace of H2O in the ATR-FTIR or 1H
NMR spectra (Figure S1; 1.1 g, 49% based on Ph3P). ATR-FTIR:
(cm−1) 3235, 1436, 1169, 1154, 1118, 1092, 1068, 1029, 996, 764,
750, 720, 692, 535, 506, 463,442. 1H NMR (MeCN-d3, 300 MHz, 298
K): δ 7.63 (30H, m, C6H5), δ 8.78 (2H, s, H2O2). 31P NMR (MeCN-
d3, 300 MHz, 298 K): δ 26.25. Caution! Although we never encountered
any explosions, care should be taken when manipulating anhydrous sources
of H2O2.
hydrogen peroxide. This latter point may indeed curtail the
scope of organometallic catalyzed water-splitting systems
without a tandem catalyst to convert H2O2 into O2 to prevent
back reactions.
Despite the disadvantages to H2O2 as a product in water
splitting, H2O is the ideal donor in photochemical hydrogen
production. Proton reduction to H2 is a two-electron process,
thereby restricting the oxidation reactions on the donor side in
traditional three-component photochemical reactors (water
oxidation to O2 is a four-electron process). To circumvent this
dilemma, Nocera has pioneered the strategy of coupling the
two-electron H2 evolution reaction with other two-electron
oxidations such as X2 formation (e.g., Cl2).10,14 Hydrogen
peroxide derived from water constitutes a suitable two-electron
redox partner to couple hydrogen evolution, but only if a
catalase is present to circumvent the back-reaction. To this end,
it is important to recognize that Milstein’s ruthenium complex
appears to behave as a dual catalyst; in one cycle it splits water,
while in another it behaves as a catalase producing O2 as a final
product. Therefore, additional efforts to understand this
organometallic catalase activity may provide the necessary
advancement to application.
Progress in this area is impeded, because there is less known
about the photochemistry of organometallic centers containing
water-derived ligands in comparison to anhydrous systems
within the established field of organometallic photochemistry.15
Thus, we are interested in probing the reactivity of organo-
metallic chromophores with water-derived ligands on earth-
abundant metal centers. We recently reported our exploration
of the formation mechanism of the tetrameric complex
[Mn(CO)3(μ3−OH)]4 (1) and briefly stated that irradiation
of this complex with UV light resulted in the formation of H2
and H2O2 along with additional products.16 Herein we describe
these findings in detail. The results indicate 1 produces H2 with
and without added water, and H2O2 forms only when water is
present. In addition to our mechanistic investigation with 1, we
screened several other L2XMn(CO)3 complexes for compara-
tive purposes (L = neutral ligand; X = anionic ligand). Certain
observations from these studies in addition to the ones with 1
help elucidate a plausible mechanism.
Photochemistry Methods. Conditions for Photochemical
Experiments. Irradiations were performed using a Newport Corpo-
ration APEX2-Xe 100 W xenon arc lamp. Irradiations of L2XMn(CO)3
complexes were conducted in degassed n-hexane/water biphasic
solutions. Because of solubility, the irradiations of 1 were conducted
in benzene or toluene rather than hexane. Various filters were
employed with the xenon arc lamp: no filter (broadband), UV band-
pass filter (50% Abs Cutoff = 382 nm; %Tavg ≥ 70% = 300−375 nm; %
Tavg ≤ 1% = 400−635 nm; %Tavg ≤ 10% = 280−290 nm), a 345 nm
cutoff filter (λ ≥ 345 nm), and a 420 nm cutoff filter (λ ≥ 420 nm).
Samples of L2XMn(CO)3 or 1 in appropriate solvent were prepared
inside of a N2-filled glovebox, and the Schlenk tube was sealed with an
unpunctured Suba·Seal septum that was secured to the flask with
electrical tape. For irradiations containing water, the water was
degassed by three freeze−pump−thaw cycles and carefully transferred
to the Schenk tube with a gastight syringe. GC headspace analysis
prior to photolyses were occasionally checked and reproducibly do not
contain O2. On the rare occasion that O2 was observed in GC traces
(either before or after a photolysis), the samples were discarded and
not included in our analyses. Samples were positioned 5 cm from the
aperture of the lamp housing, and the stir rate was always the same for
each reaction. For gas quantification, a sample of gas (3 mL) was
obtained with a 10 mL gastight syringe; the puncture was sealed with a
dollop of grease if subsequent sampling was desired. For quantification
of organic products, the flask was open to the air, and 500 μL of the
organic phase was removed and diluted with 500 μL of hexane
EXPERIMENTAL SECTION
■
Physical Methods. NMR experiments were performed on Varian
Mercury 300 MHz, Inova 400 MHz, and Inova 500 MHz
spectrometers. Headspace analysis was obtained using a PerkinElmer
Clarus 580 GC (thermal conductivity detector, Ar carrier gas). Gas
chromatography-mass spectrometry (GCMS) analysis of reaction
mixtures was performed using an HP 5890 Series II GC coupled with a
HP 5972 Series Mass Selective Detector. Mass spectral analysis of
isotopically labeled dihydrogen was performed on a Thermo Finnigan
MAT95 XL high-resolution mass spectrometer equipped with an
electron impact ionizer. Transmission and attenuated total reflectance
(ATR)-Fourier transform infrared (FTIR) spectra were collected
inside of a VAC Atmospheres Omni glovebox using a Bruker Alpha IR
spectrometer with ALPHA-T Universal sampling module and ALPHA-
P Platinum ATR module (diamond crystal), respectively. UV−vis
spectra were collected using an 8154 Agilent Spectrophotometer.
CHN combustion analyses were performed by Robertson Microlit
Laboratories, NJ. X-band electron paramagnetic resonance (EPR)
spectra were collected using a Bruker EMX EPR spectrometer.
Synthetic Methods. All reagents were obtained from commercial
vendors and used as received unless otherwise noted. Unless otherwise
stated, all manipulations were performed in a nitrogen-filled Genesis
VAC glovebox or under argon using typical Schlenk line techniques.
Solvents were dried and obtained from a PPT solvent purification
B
DOI: 10.1021/acs.inorgchem.7b01483
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
containing 0.001 M dodecane. Calibration curves were prepared for
each analyte (e.g., see Figure S2).
Observation of [CptetMn(CO)2]2(μ-CO) from Photolysis of
CptetMn(CO)3. CptetMn(CO)3 (0.0212 g, 0.08 mmol) was dissolved
in 5 mL of dry benzene and photolyzed for 1 h under an argon
atmosphere to obtain a green solution (the same green species is
generated in wet or dry hydrocarbon solvent). Headspace analysis by
GC confirmed 0.5 equiv of CO. In situ UV−vis monitoring was
performed in a gastight cuvette. Removal of the solvent in vacuo
resulted in a green oil. UV−vis (in situ or isolated oil in benzene): λmax
(benzene, nm) 320, 450, 680. ATR-FTIR (thin film, cm−1): 2002,
1904, 1772. In situ monitoring with 1H NMR: 0.005 g of
CptetMn(CO)3 was photolyzed in C6D6 for 2 h in J-young NMR
Characterization of White Solid (2) Obtained from Irradiations in
Dry Toluene. The white precipitate (2) formed during an overnight
photolysis of 1·(C6H6)2 (100 mg, 0.13 mmol) in 30 mL of toluene was
isolated on a fritted glass funnel inside of a nitrogen-filled glovebox
(exposure of this solid to air results in rapid color change to dark
brown) and subsequently washed with toluene and dried in vacuo (40
mg). The ATR-FTIR spectrum of the solid was obtained inside of a
glovebox: (IR(cm−1)) ν(OH) = 3650, ν(OD) = 2700, ν(CO) = 1850
(Figure S3). CHN analysis Anal. Calcd (found) for
[Mn3(CO)2(OH)3(O)]: %C 8.35 (8.17); %H 1.05 (1.14); %N <
0.02 (0). EPR samples of the pure material and 1% in NaHCO3 were
prepared in the glovebox and covered in mineral oil.
Conversion of Hydrogen Peroxide into Dioxygen after Irradiation
of 1 in Wet Solvents. Irradiation of compound 1·(C6H6)2 (3.9 mM) in
a biphasic solution of benzene/water or toluene/water (9:1 mL) in a
sealed 25 mL Pyrex Schlenk tube using the 300−375 nm band-pass
filter for 5 min resulted in the formation of H2O2 and a white
precipitate suspended in the aqueous layer. After irradiation, 1 mL of
degassed sodium phosphate buffer solution (pH 7, 0.1M) was added
to the reaction mixture through the septum with a syringe, followed by
removal of 10 mL of headspace gas to depressurize the closed system;
it is imperative to exclude oxygen from these experiments. A solution
of catalase (1 mg/10 mL) in sodium phosphate buffer (pH 7, 0.1M)
was degassed, and addition of 200 μL of this solution directly into the
aqueous layer using a long needle resulted in formation of tiny bubbles
from the aqueous layerit is critical that the catalase not come into
contact with the organic phase, as the catalase immediately precipitates
if contact is made. The reaction mixture was allowed to stir at a very
low rate for 10 min before analyzing the headspace for O2. The
headspace was analyzed before and after the addition of catalase; O2
was only observed after bubble formation upon catalase addition. The
headspace was also analyzed by GCMS when 18O-1 isotopologues
were used and 34O2 analyzed for.
1
tube. H NMR (400 MHz, C6D6, ppm): δ 3.68 (s, 1H, CMe4H), δ
1.43 & δ 1.38 (s, s, 6H, 6H, CMe4H) for CptetMn(CO)3; δ 4.22 (s, 1H,
CMe4H), δ 1.67 & δ 1.59 (s, s, 6H, 6H, CMe4H) for [CptetMn-
(CO)2]2(μ-CO). The UV−vis and FTIR data of the green species is
tentatively assigned as [CptetMn(CO)2]2(μ-CO) due to similarity with
the previously described FTIR and UV−vis data for [CpMn-
(CO)2]2(μ-CO).24
Observation of Mn2(bipy)2(CO)6 from Photolysis of (bipy)Mn-
(CO)3Br. A 30 min irradiation of (bipy)Mn(CO)3Br (17 mg, 0.05
mmol) in wet or dry hydrocarbon solvent resulted in the formation of
a purple solution (the purple solution decolorized upon exposure to
water or air). When the irradiation was performed in dry solvent, it
was possible to isolate a purple solid by removal of the volatiles in
vacuo. UV−vis, λmax (benzene, nm): 294, 853. ATR-FTIR: (cm−1)
3101, 3069, 3024, 2022, 1953, 1913, 1888, 1860, 1836. The data
matches those for the known dimeric species Mn2(bipy)2(CO)6.25
Observation of MnTp2 from Photolysis of TpMn(CO)3. A 5 mL
toluene solution containing 0.012 M TpMn(CO)3 (21 mg, 0.06
mmol) was treated with 100 μL of H2O and degassed with three
freeze−pump−thaw cycles. The solution was subjected to 1 h of
broadband irradiation, and water was removed in vacuo; the colorless
solution was filtered and analyzed with thin-film ATR-FTIR. ATR-
FTIR: (cm−1) 2468 ν(BH), 1500, 1400, 1388, 1297, 1202, 1111, 1046,
971, 875, 756, 716, 656, 616. These features are identical to those
published for MnTp2.26
Actinometry. The intensity of the irradiating light (I0) just outside
the front window of the photolysis cell was measured by a chemical
actinometry method employing potassium ferrioxalate (K3Fe(C2O4)3·
3H2O) to calculate the quantum yield.27 A solution of 0.006 M
K3Fe(C2O4)3·3H2O in 1.0 N H2SO4 was prepared, and 10 mL of this
solution was used for each irradiation. Irradiation conditions such as
the reaction flask, distance from the aperture of the lamp, filter, and stir
rate were identical to the standard photolysis conditions used for
irradiation experiments of compound 1. The irradiations were done for
5, 10, and 15 s. Then 1 mL of photolyzed solution (V1) was mixed
with 0.5 mL of HOAc (0.6 N, pH 4.6) and 2.0 mL of 0.1% 1,10-
phenanthroline and then diluted to 10 mL (V) with water. An identical
sample with unphotolyzed solution was also prepared and used as the
blank for UV−vis. The mixed solutions were then allowed to stand for
1 h in the dark before analysis by UV−vis. The absorption spectra
were collected, and the absorbance at 510 nm of the solution was
measured. All preparation and manipulation of the ferrioxalate solution
were performed in a dark room, using a red photographic safelight.
Finally, the light intensity of the examined light source (I0) was
determined by eqs 1 and 2, where A is absorbance, V is the final
volume (10 mL), V1 is the volume of irradiated actinometer solution
used to prepare V (1 mL), ε is extinction coefficient (ε = 11 100 M−1
cm−1), t is irradiation time in seconds, ΦFe(II) = 1.28, and f = 1. The
final I0 value (1.69 0.05 × 10−6 einstein s−1) was determined by
averaging the I0 for all three irradiation times. This process was
performed in duplicate.
Synthesis of Ph3P-Oxide by Reacting Ph3P with Photogenerated
H2O2 from 1. Irradiation of compound 1·(C6H6)2 (1.3 mM) in a
biphasic solution of toluene/water (9:1 mL) in a sealed 25 mL Pyrex
Schlenk tube using the band-pass filter for 5 min resulted in the
formation of H2O2 and a white precipitate suspended in the aqueous
layer. The liquid portions were separated from solids via Schlenk
filtration into a flask containing premassed Ph3P and allowed to stir
overnight. The sample was then lyophilized to remove volatiles.
GCMS samples were prepared in the glovebox under a nitrogen
atmosphere and subsequently analyzed.
Hydrogen Peroxide Quantification. The technique used here for
hydrogen peroxide detection is a modified version of published
methods.22 A reaction mixture containing a known amount of 1 (∼1
mM in 10 mL of dry toluene) was subjected to UV irradiation for a
specified time. After photolysis an equal volume of water (10 mL) was
added to the reaction mixture under sealed conditions via a syringe.
After it was stirred for 5 min, the aqueous portion was separated. This
aqueous portion (5 mL) was then treated with equal volume (5 mL)
of 1 M KI solution in 0.05 M H2SO4, and the solution was allowed to
rest for 5 min. A 1 mL aliquot was diluted with water to 10 mL and
directly analyzed by UV−vis spectroscopy in a 1 cm quartz cuvette.
After 5 min the absorption of I3− at 353 nm (ε = 26 000 M−1 cm−1)23
was determined, and the yield of H2O2 from photolysis experiment
was calculated (Figure S4). Ammonium molybdate (two drops, 3%
−
aqueous) can be added to accelerate the formation of I3 and was
necessary when concentrations of H2O2 were very low. Addition of
ammonium molybdate was not necessary in experiments with 1, as the
−
concentration of H2O2 was sufficiently high for I3 to fully develop
within the first 5 min. When photolyses of 1 were conducted in the
presence of water (1 mM 1 in 9 mL of toluene +1 mL of water), 9 mL
of water was added to the mixture instead of 10 mL and subsequently
V × 10 × A
moles of Fe(II) =
1000 × ε × V
(1)
(2)
1
−
worked up as describe above. Note: the I3 will continue to develop
over time due to a slow background oxidation from air; therefore, it is
important to ensure that samples do not develop for much longer than
5 min.
moles of Fe(II)
I0 =
ϕFe(II) × t × f
C
DOI: 10.1021/acs.inorgchem.7b01483
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
a
Table 1. Yields of H2 and H2O2 Produced from 1 h of Irradiations of L2XMn(CO)3 Complexes
b
c
c
entry
complex
conditions
light source
molar equiv of CO
% yield H2
% yield H2O2
1
1
1
1
1
1
toluene/H2O (9:1)
300−375 nm
300−375 nm
300−375 nm
300−375 nm
300−375 nm
300−375 nm
broadband
8.0
4.2
7.0
7.7
4.0
2.0
2.8
0.67
2.7
2.9
2.9
0
1
1
1
79 14
22
40
10
0
2
4
1
d
2
toluene/H2O (9:1)
28
84
45
24
21
24
03
33
29
33
0
5
5
e
3
benzene/H2O (9:1)
f
4
toluene
f
5
benzene
0
6
CpMn(CO)3
CpMn(CO)3
Mn2(CO)10
toluene/Η2O (9:1)
hexane/H2O (5:1)
toluene/H2O (9:1)
hexane/H2O (5:1)
hexane/Η2O (5:1)
toluene/Η2O (9:1)
0.8
12
0
7
8
300−375 nm
broadband
9
Mn2(CO)10
0
10
11
12
[TpMn(CO)3]
(bipy)Mn(CO)3Br
no complex
broadband
0.5
0.8
0
300−375 nm
broadband
hexane/H2O (5:1)
a
b
See Tables S1 and S2 for additional data. Irradiations were conducted in Pyrex glass Schlenk flasks using a Newport Corporation APEX2-Xe 100
W xenon arc lamp as the light source: 300−375 nm = UV-band-pass filter (%Tavg ≥ 70% = 300−375 nm), broadband = no filter, λ ≥ 345 nm = 345
c
d
e
f
nm cutoff filter, λ ≥ 420 nm = 420 nm cutoff filter. Entries 6−12 did not produce H2 when water was excluded. 5 min. 120 min. Solvent was
dried with freshly activated alumina.
Quantum yield of product (Φp) was determined using eq 3, where P is
the product:
Scheme 1. Notable Observations from Photochemical
Experiments with Monomeric L2XMn(CO)3 Complexes
mol P
I0 × t × f
ϕ =
p
(3)
RESULTS
■
Photochemical Water-Splitting Studies with L2XMn-
(CO)3 Complexes. We explored a variety of manganese
carbonyl compounds within the context of Fan’s original
discovery of water splitting with cymantrene and hydrogen
production from water with Mn2(CO)10 for comparative
purposes with 1 (Table 1, entries 6−9).9,28 Generally, our
results are consistent with those found by Fan and co-workers.
Solvents such as MeCN, tetrahydrofuran (THF), benzene, and
toluene retard H2 formation, and a majority of the compounds
studied here produce H2O2 in yields below 1% regardless of
solvent. In the few cases that did produce appreciable yields of
H2O2 (12% Table 1, entry 7), only broadband irradiation was
sufficient. Despite the intriguing reaction mediated by
cymantrene, the overall water-splitting chemistry with the
L2XMn(CO)3 complexes is unimpressive, and there are no
obvious trends from varying the ancillary ligand (Table S2).
Although not the focus of this report, we briefly describe
some notable observations from studies with complexes other
than 1, because they aid in the interpretation of the
photochemical water-splitting reaction mediated by 1 (Scheme
1 and Figure S5). For example, irradiation of CptetMn(CO)3 in
wet or dry hydrocarbon solvent produced 0.5 equiv of CO and
a metastable green species with spectroscopic features strikingly
similar to the known dimer [CpMn(CO)2]2(μ-CO).24 Use of
Cp*Mn(CO)3 gave essentially the same result and in both
cases produced almost no hydrogen. Thus, we propose that
photolysis of CptetMn(CO)3 and Cp*Mn(CO)3 forms the
dimeric [CpRMn(CO)2]2(μ-CO) complex and that the absence
of H2 suggests that water binding is too slow for productive
water-splitting chemistry, perhaps due to steric effects. The only
tractable product we obtained from the photolysis of
(bipy)Mn(CO)3Br in wet or dry hydrocarbon solvent was
the dimer [(bipy)Mn0(CO)3]2 (Table 1, entry 11); we
obtained ∼30% H2 when water was present in photolyses.
Irradiation of TpMn(CO)3 in wet hydrocarbon solvent
produced the MnIITp2 species along with H2 and 3 equiv of
CO (Table 1, entry 10). The presence of Mn0 and MnII species
after photolysis of the L2XMnI(CO)3 complexes indicates that a
disproportionation reaction might be operative and is
consistent with observations made in the photolysis of 1 and
[MnI(CO)3(OH2)3]+ (vide infra).
Photochemical Decomposition of 1. While hexane/
water mixtures proved the best for producing H2 with the
L2XMn(CO)3 compounds, complex 1 is insoluble in hexane
and is only sparingly soluble in benzene. Also note that 1 is
completely insoluble in water. Therefore, we conducted the
majority of our photochemical experiments with 1 in dry or
biphasic water/toluene solutions; a recent investigation
discusses some of the advantages to using biphasic mixtures
in photochemical water-splitting reactions.29 Our interest in 1
stems not only from the fact that 1 already contains water-
derived ligands and produced the highest yields of H2 and
H2O2 compared to the L2XMn(CO)3 complexes but also
because of its interesting structure. Moreover, our broadly
D
DOI: 10.1021/acs.inorgchem.7b01483
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
stated hypothesis that organometallic chromophores with
water-derived ligands mediate photochemical water-splitting
reactions is testable with 1. Therefore, we sought to obtain
enough data to construct a mass-balanced photochemical
reaction for mechanistic purposes.
Broadband irradiation of a homogeneous solution of 0.001 M
1 in benzene/water or toluene/water biphasic mixture in a
gastight Pyrex Schlenk tube with a Xe-arc lamp causes
immediate formation of an off-white precipitate, carbon
monoxide, dihydrogen gas, 0.5% o- and 0.3% p-methyldiphenyl-
methane (in the case of toluene only), and H2O2 (Schemes 2
Scheme 2. Summary of Products Obtained from a 5 min UV
Irradiation of 1 in Anhydrous Toluene
and 4). No set of conditions produced hydroxylated organic
products or CO2. We obtained different yields of H2O2 and H2
depending on the filter used and irradiation times (Table 2).
Table 2. Effect of Irradiation Time and Filters on H2 and CO
a
Yields with 1
Figure 1. Yield of gaseous products from photolysis of 1 as a function
of time in toluene/H2O 9:1 (top) and dry toluene (bottom) using the
300−375 nm filter.
b
c
filter
% H2
% CO
% H2O2
ΦCO
t (s)
300−375
300−375
300−375
300−375
300−375
≥420
0.2
0.4
1.0
23
39
78
0.23
0.20
0.20
0.14
0.12
0.13
0.14
0.16
13
30
158
220
480
520
540
630
60
identification and is an ongoing effort in our laboratory.
Nevertheless, a combination of FTIR, EPR, and CHN analysis
proved informative and beneficial for mass-balance purposes.
For example, the ATR-FTIR spectrum of 2 reveals the presence
of OH (ν(OH) = 3650 cm−1) and metallo-carbonyl (ν(CO) =
1850 cm−1) groups (Figure S3). The former is derived from the
μ3−OH moieties in 1 evidenced by the shift to 2700 cm−1
when d4-1 is used for photolysis. An EPR spectrum of 2 at
room temperature indicates that this material contains
paramagnetic centers, but they lack hyperfine structure even
if diluted 100-fold in NaHCO3 (Figure S3). A CHN analysis of
2 reveals an empirical composition consistent with [Mn3(μ-
CO)2(OH)3(O)], which essentially corresponds to 1 minus a
H atom and a Mn0 center and 10 CO ligands; however, this
assignment is only one possibility, as it could just as easily be a
mixture of various compounds. As will be discussed below, both
H atoms and Mn0 are present at various stages of the
photochemical decomposition reaction.
We speculate that 2 is an arrested manganese(I/II)-carbonyl
material en route to a similarly ill-defined MnII(bi)carbonate
species that is observed when water is included in photo-
chemical reactions with 1 or L2XMn(CO)3. For instance, when
2 is treated with air-free water it produces a different material
with an ATR-FTIR spectrum closely resembling the white
material that is obtained when 1, cymantrene, or Mn2(CO)10 is
irradiated in hydrocarbon/water biphasic mixtures; these ill-
defined white species are not the same, but they all contain
stretches in the region associated with inorganic carbonates and
hydroxides moieties (Figure S3).30 Fan and co-workers
28
120
300
300
300
300
33
40%
2
4
18
≥345
23
2
d
none
35
34
a
b
Conditions: toluene/water (9:1), 0.001 M 1. See text for details on
filter. See actinometry in Experimental Section, quantum yields for
other products are 2 orders of magnitude smaller (e.g., ≤0.006). No
filter = broadband irradiation.
c
d
The highest yield of H2O2 is produced with a UV band-pass
filter (%Tavg ≥ 70% = 300−375 nm) and unless otherwise
stated was used for most the experiments described below. The
yield of dihydrogen produced reaches as high as 79%
14
relative to the initial concentration of 1. A time course (Figure
1 and Table S1) reveals that H2 forms slowly, indicating that it
is not arising directly from the primary photochemical reaction
(ΦH2 = 0.006 at 5 min). Moreover, dihydrogen and CO do not
continue to form in the dark. The ratio of carbon monoxide
maintains 8:1 over 60 min of irradiation with an end point of 8
mol equiv (765% 98). The yields of CO vary from overnight
photolyses but often exceed 11 equiv.
Characterization of the Precipitates. Some of the carbon is
sequestered by the manganese in the form of MnII(bi)-
carbonate salts as indicated from the EPR and ATR-FTIR
characterization of the precipitates. However, if water is
excluded, a different “anhydrous” precipitate forms (2). The
rapidity at which 2 decomposes in air has complicated its
E
DOI: 10.1021/acs.inorgchem.7b01483
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
described the solids as Mn(CO3),28 but we believe they are
complicated mixtures of ill-defined manganese salts that
possibly contain OH and/or CO3 groups.
Scheme 3. Scrambling of d4-1 Deuterons with Water and
Aromatic Solvent
Investigating the Source of H2 in Irradiation of 1. Unlike
the other complexes used in Table 1, complex 1 does not need
added water for H2 production (Scheme 2). For instance, if
water is rigorously excluded from irradiations in benzene or
toluene H2 is produced in 24% and 45% yields, respectively.
Higher yields of H2 can be obtained for longer irradiation times
and are affected by different solvents such as THF and
MeCN.31 No HD or D2 is observed by mass spectrometry
when 1 is irradiated in d8-toluene. We attempted similar
experiments with d4-1 and mixtures of 1 and d4-1 in
nondeuterated solvents and obtained inconsistent ratios of
H2, D2, and HD (Figure 2). We note that during the
water despite our best efforts to exclude it from certain
experiments in addition to scrambling with toluene solvent.
After the first 60 s, the quantum yield of CO begins to drop
substantially. At this stage, photons are no longer a limiting
reagent, and the reaction mechanism can no longer be reliably
interpreted. Nevertheless, the continued production of gaseous
products during prolonged irradiation suggests to us that after
initial loss of CO (forming 1-CO), secondary photochemical
and/or thermal processes begin (sustained irradiation is
required to reach the maximum yields of CO, H2O2, 2, and
H2). We therefore probed the solution contents after a 10 min
broadband irradiation to gain insight into these processes and
found that 1 is completely consumed and Mn2(CO)10 is the
only manganese-containing soluble compound obtained in
8.1%
1.6 yield (Figure S7). The formation of Mn2(CO)10
upon photolysis of 1 is not so unlike the results obtained by
Bamford, who found that Mn2(CO)10 is produced upon
photolysis of [MnI(CO)3(OH2)3]+ in aqueous solutions and
proposed a disproportionation reaction.33 Moreover, we
obtained Mn0 and MnII compounds in the photolysis of
(bipy)Mn(CO)3 and TpMn(CO)3, respectively, as described
above. Although it may only be a minor component of the
reaction, a disproportionation of 1-CO accounts for our
observation of Mn0 compounds and the presence of para-
magnetic centers in 2 that form in the photolysis of 1 in
anhydrous conditions.
The presence of Mn2(CO)10 from photolysis of 1 caused us
to consider the possibility that Mn2(CO)10, and not 1, was
responsible for the formation of H2 and H2O2. However, no H2
is produced when Mn2(CO)10 is irradiated in dry solvents, nor
does H2O2 form when Mn2(CO)10 is irradiated in wet solvent
thus confirming that species derived uniquely from 1 are
responsible for the observed photochemical products. We did,
however, confirm Fan and co-workers findings that H2 is
produced by the photochemical reaction between Mn2(CO)10
and water in hexane (30% yield, entry 9, Table 1).28 However,
the yield of H2 in wet toluene is small (3% yield, entry 8, Table
1). In another control reaction we demonstrated that
Mn2(CO)10 and anhydrous [Ph3PO]2·H2O2 in benzene or
toluene produce H2 in a photochemical reaction (31% yield)
providing an explanation for the steady decrease of H2O2 in
photochemical reactions with 1 despite the stability of
[Ph3PO]2·H2O2 under irradiation by itself (vide infra). In
conclusion, the presence of Mn2(CO)10 shows that Mn0 species
form during the photolysis of 1 and implicates a possible third
process for H2 production, namely, a photochemical process
that involves H2Ox (x = 1,2) and Mn0 species derived from 1.
Investigating the Source of Oxygen in H2O2 from
Irradiation of 1. Following H2O2 yield as a function of time in
photolyses of 1 with water indicates that it rapidly increases at
first and then steadily decreases as the yield of H2 increases
(Figure 1). Samples analyzed after 5 min of irradiation of 1 in
Figure 2. Mass spectra of dihydrogen produced from photolysis of 1
and d4-1 in dry toluene or d8-toluene. The two bottom spectra are
representative examples of the same experiment repeated with
different but identically prepared samples demonstrating the difficulty
in reproducing isotopic ratios when d4-1 was used.
preparation of d4-1 it is imperative to use solvent that has
been dried with freshly activated alumina; otherwise, the ATR-
FTIR spectrum of d4-1 contain peaks associated with unlabeled
1. Furthermore, it appears that d4-1 scrambles deuterium with
protons of aromatic solvents like toluene. For example, a
ν(OH) slowly appears in the transmission FTIR spectrum
when d4-1 dissolved in toluene dried with alumina (Figure S6).
2
Additionally, H NMR spectroscopy reveals the presence of
deuterated toluene when d4-1 is dissolved in toluene. It is
known that 1 forms H-bonds with aromatic solvent molecules
and may facilitate such scrambling (Scheme 3).32 Taking all this
into consideration, we posit that the inconsistent ratio of
labeled dihydrogen isotopes arises from scrambling with trace
F
DOI: 10.1021/acs.inorgchem.7b01483
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
rigorously dried toluene or benzene contain only trace amounts
of H2O2 (≪1%). For these experiments, the solvent was dried
by passing it through a column of freshly activated alumina. If
the freshly activated alumina was not used to dry the solvent
prior to photolysis, hydrogen peroxide was detected in slightly
higher yields in benzene or toluene (max 2%, ΦH2O2 = 0.0013, t
= 5 min). If water is purposefully added, the yield of hydrogen
peroxide jumps to 40% 4 (max 45%, ΦH2O2 = 0.007, t = 5
min). We also emphasize that the yields of H2O2 are consistent;
irregular yields might indicate O2 leakage and subsequent
reduction to H2O2.
and [Ph3PO]2·H2O2 was photolyzed in the absence of 1. This
experiment confirms that we are not irradiating with photons
energetic enough to homolyze H2O2.34
When DMPO and 1 were irradiated together in dry toluene,
EPR spectroscopy confirmed the presence of two radicals
consistent with the {DMPO-benzyl}· and {DMPO-H}· radicals
(Figure 3).35,36 When benzene was used, the {DMPO-H}· was
To determine the source of O atoms in the H2O2 product we
analyzed O2 and phosphine oxides generated from the
synthesized H2O2. For instance, anaerobic treatment of
postirradiation aqueous portions with catalase liberates O2
from the H2O2 generated from water in photolyses with 1.
GCMS of the gas when 18O isotopologues of 1 is used includes
an m/z for 34O2 with intensity only slightly above the
background of naturally occurring 34O2. Similarly, when the
photogenerated hydrogen peroxide is used to prepare Ph3PO,
only trace label incorporation from 18O-1 is achieved. Taken
together, these experiments show that water is necessary for the
formation of H2O2 and that the hydroxido oxygen atoms in 1
are not predominantly incorporated into product (Scheme 4).
Scheme 4. Summary of Products From a 5 min UV
Irradiation of 1 in Wet Toluene and Subsequent Reactions
with Water-Derived H2O2
Radical Trapping Experiments with 1. Methane is a
product when 1 is irradiated in MeCN (16%) and o- and p-
methyldiphenylmethane (<1%) in toluene. Although difficult to
reproduce, biphenyl was observed in some cases when 1 was
irradiated in benzene. Despite the low yields of these products,
their presence suggested possible radical processes in the
photochemical decomposition of 1. Moreover, we wanted to
test for hydroxyl radicals as a possible source of the H2O2.
When (2,2,6,6-tetramethylpiperidin-1-yl)oxidanyl (TEMPO),
5,5-dimethyl-1-pyrroline N-oxide (DMPO), or Ph3P is included
in photochemical reactions with 1, the production of H2 is
completely suppressed. Of these reagents, only TEMPO is a
radical, and thus the suppression of H2 production may arise
from a coordination of the reagent after CO loss with a rate
that is competitive for the process that produces H2. In fact,
various soluble manganese−carbonyl compounds are produced
when TEMPO and Ph3P are included, but these compounds
have not been characterized beyond ATR-FTIR spectroscopy.
When DMPO was added to a solution of 1 in dry toluene
after a 5 min photolysis, no radical species were observed via
EPR spectroscopy except for those associated with the
precipitates (vide supra). Thus, there are no long-lived radicals
that form during the irradiation experiments. Additionally, no
radicals were observed when an anhydrous solution of DMPO
Figure 3. RT X-band EPR spectra (black) were collected immediately
after a 5 min photolysis of 1 in toluene (top) or benzene (bottom) in
the presence of 4−5 molar equiv of DMPO. See Table S3 for
simulation (red) parameters.
the major radical species with a minor radical component
consistent with {DMPO-phenyl}·. For the irradiations in
anhydrous solvent, there was no indication of {DMPO−
OH}· or {DMPO-OOH}· radicals. The EPR spectra obtained
from irradiations with 1 and DMPO in toluene/water were
essentially silent with radicals barely perceptible above the
baseline. Unfortunately the noise level is too great to accurately
model the radicals generated in wet solvent, and their identity
remains uncertain.
DISCUSSION
■
The water-splitting systems studied by Milstein and Beweries
contain water-derived ligands prior to photon absorption.
Additionally, the O−O bond-forming step is the photochemical
step. Compare this to water splitting with cymantrene, which
involves photochemical ligand dissociation prior to substrate
G
DOI: 10.1021/acs.inorgchem.7b01483
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
water binding. Currently, the proposed mechanism for
cymantrene water splitting is purely thermal after initial
photochemical loss of CO. As unlikely it is that endergonic
catalytic water splitting can be driven thermally, we confirmed
Fan’s observation of photochemical H2 and H2O2 formation
using cymantrene and Mn2(CO)10 in addition to using several
other L2XMn(CO)3 variants. An immediate conclusion from
our photochemical reactions (Table 1 and Table S2) is that the
water-splitting reaction is affected by supporting ligand, but
there are no obvious trends. The highest yields of H2 and H2O2
are from irradiations of 1. One potential reason for the higher
overall yield is that water-derived ligands in 1 are present prior
to photolysis in contrast to the other manganese complexes
described above, and this hypothesis guided our efforts toward
understanding the photo-decomposition of 1.
Scheme 6. Proposed Mechanism for Photochemical Water
Splitting Mediated by 1
Hydrogen Peroxide Quantification Methodology. One
of our primary goals was to understand the mechanism of
photochemical water splitting using 1, which thus necessitated a
reliable methodology to quantify H2O2 generated under the
conditions studied here. The need arose specifically because
interpretation of results from the traditional methods were
complicated by the fact that manganese-containing compounds
after photolysis interfered with the absorption of colorimetric
indicators giving false positives and erroneously high yields.37
Of the several methodologies we considered, the most
attractive was quantitative iodometric determination with
UV−vis spectrophotometry.22,23 Use of this method enables
reliable determination of H2O2 concentration in aqueous
solutions prepared with 30% aqueous H2O2.
This premise is supported by the presence of Mn2(CO)10
and paramagnetic MnII centers in 2 that form under anhydrous
conditions. Furthermore, the stoichiometry of 2,
[Mn3(CO)2(OH)3(O)], is such that all oxygen atoms
originating from the hydroxido ligands in 1 are accounted for
along with the three manganese centers. The dihydrogen and
simultaneous precipitation of 2 is thus proposed to form by
bimolecular coupling of 3. In part, this is supported by the
presence of HD, D2, and H2 in experiments where d4-1 was
irradiated in the presence of 1. Additionally, the formation of
{DMPO-H}· instead of H2 when DMPO is included in
photolysis further supports this hypothesis. While at this stage
our mechanistic proposal for the formation of H2 via 3 is only
speculative, there is ample evidence that indicates a species
uniquely derived from 1 forms dihydrogen in a thermal (kH2) or
photochemical step (hν2) under anhydrous conditions. Finally,
since solvent is not directly involved in the H−H forming step
and the hydroxide ligands in 1 are derived from water, this H2
production reaction represents the reduction half reaction in
water splitting by 1.
However, quantification of H2O2 generated in organic
solvents, such as toluene, required additional validation. For
this, we synthesized an anhydrous form of H2O2 following the
lead of Blumel and co-workers.21 They prepared various
̈
hydrated phosphine oxide−hydrogen peroxide adducts includ-
ing [Ph3PO]·H2O2·H2O, from which we modified their
procedure to obtain the anhydrous [Ph3PO]2·H2O2 form
(Scheme 5). Thus, our method was also validated for
The oxidative side of water splitting occurs only when water
is included. For example, the H2O2 product appears to originate
from free water, since label incorporation from 18O-1 is not
found in the oxidized product. This was demonstrated by
analyzing O2 liberated from the synthesized H2O2 with catalase
by GCMS or using the H2O2 to generate phosphine oxides.
Moreover, H2O2 is only formed when water is included in
photolyses. Since the complex 1 does not exchange with free
H218O in MeCN, water must interact (k2/k−2) with 1-CO or a
species resulting from it, such as 3.16 This proposal is
reasonable considering the rapid exchange of water with the
known complex [MnI(CO)3(OH2)3]+, which is among the
fastest for d6 transition-metal carbonyl aqua compounds.38
Another convincing argument in support of H2O2 forming
from interaction of water with 1-CO or 3 is that the photons
used in this experiment are not energetic enough to split water
alone. Specifically, to photochemically split water directly one
requires more than 6.5 eV photons;39 it is unlikely that such a
process is occurring given that our lamp provides less energetic
photons less than 4.4 eV. While this is not energetic enough to
homolyze water via direct sensitization, the various filters we
used in photochemical decomposition of 1 have the largest
impact on H2O2 yield; H2 and CO are produced regardless of
filter, but H2O2 is only produced when λ < 345 nm is used
(Table 2). Therefore, the amount of energy provided by the
lamp directly influences the water-splitting reaction, the
Scheme 5. Synthesis of an Anhydrous Source of H2O2
quantifying H2O2 extracted from toluene solutions containing
a known amount of pure [Ph3PO]2·H2O2; subsequent analysis
of the aqueous portion revealed that 91% ( 3) H2O2 was
extracted into the water.
Proposed Mechanism for H2O2 and H2 Production
from 1. The light-driven formation of H2 and H2O2 mediated
with 1 is complicated as evident from our investigation.
Reflecting upon the results from photo-decomposition of 1, we
cautiously speculate a mechanism (Scheme 6). First, on the
basis of the high quantum yield of CO compared to other
products, the initial photochemical step (hν) is loss of CO
ligand, forming 1-CO in the process. Species 1-CO may react
with CO (k−1) but most likely is carried forward by means of
relaxing from hot vibrational states. The result of this relaxation
is uncertain and could result in complete destruction of the
cluster or loss of several CO ligands. More likely it results in a
disproportionation (kdisprop) into a [(MnI)2MnII] species (3)
and Mn0 species.
H
DOI: 10.1021/acs.inorgchem.7b01483
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
oxidation half-reaction requiring the most energy, but it is not
directly splitting water.
To understand this phenomenon in more detail, we
investigated the electronic structure of 1. The UV−vis
spectrum of 1 in MeCN and toluene contains an intense
band at 388 nm (4800 M−1 cm−1) and 392 nm (4600 M−1
cm−1), respectively (Figure 4).16 Density functional theory
involved in the oxidative half-reaction for water splitting in a
rapid chemical (k3) or photochemical (hν3) step. We
considered the possibility that compounds similar to the
known cation [MnI(CO)3(OH2)3]+ might form upon inter-
action of 1-CO or 3 and water and subsequently promote water
splitting. To test this, we independently synthesized and
photolyzed [Mn(CO)3(OH2)3]+ but found that no dihydrogen
or hydrogen peroxide was formed (Scheme 7). However,
Scheme 7. Summary of Acid−Base Chemistry and
33,38
Photochemistry of [Mn(CO)3(OH2)3]+
[MnI(CO)3(OH2)3]+ does not contain hydroxido ions that are
likely present in 1-CO or 3. The [MnI(CO)3(OH2)3]+ ion is
known to be unstable toward base (decomposing into 1), so it
remains for a future endeavor to test these water-soluble
manganese carbonyl compounds for water splitting.38
Finally, an additional process forms dihydrogen (hν4) that
involves the interaction of H2Ox (x = 1,2) and Mn0 species
derived from the disproportionation of 1-CO that are
reminiscent of, if not identical to, the photochemical reactions
between Mn2(CO)10 and H2O or Mn2(CO)10 and [Ph3PO]2·
H2O2 (Scheme 8). In addition to the known spontaneous
Figure 4. (top) UV−vis spectrum of 1 in MeCN with overlaid
TDDFT predicted vertical transitions (solid red vertical lines).
(DFT) calculations of 1 reveal a Kohn−Sham orbital splitting
that closely resembles the cluster orbitals in other organo-
metallic chalcogenide cubane complexes with a triply
degenerate lowest unoccupied molecular orbital (LUMO)
being the last three of the thirty-two cluster orbitals (Figure
S8 and Table S4).40 For discussion purposes, these cluster
orbitals serve as the ligand-field equivalent of monomeric
organometallics. We performed time-dependent (TD) DFT
calculations and obtained three allowed vertical transitions that
agree with the experimental UV−vis features, the most intense
being 3.01 eV and the most energetic being at 4.1 eV.
Historically, these transitions were attributed to ligand-field
dπ−σ* to explain photodissociation of CO ligands from
organometallic carbonyl compounds.15 However, the bands in
the UV−vis spectrum of 1 shift in different solvents and are
likely cluster orbital to COπ* charge transfer in nature, which is
similar to the forbidden dπ−COπ* metal-to-ligand charge
transfer (MLCT) transitions in monomeric carbonyl com-
pounds as described in contemporary literature.41 Different
excitation energy may have varying CO dissociation outcomes
and affect subsequent water-splitting reactions. Such a
phenomenon has been observed with Fe(CO)5 in gas phase
and solution whereby a different number of CO ligands are
dissociated depending on the energy of photons provided,
despite the fact that only one photon is absorbed.42 Hence, the
higher-energy photons used to drive the water-oxidation half-
reaction mediated by 1 may enable the dissociation of multiple
carbonyl ligands from 1 within a time frame appropriate for
H2O2 to form.
Scheme 8. Photochemical H2 Production From Mn2(CO)10
reactions between H2O2 and Mn2(CO)10,44 the latter process
partially accounts for the disappearance of H2O2 in extended
irradiations of 1, since [Ph3PO]2·H2O2 is not decomposed into
OH· by the light source. Albeit, the formation of Mn0 species is
only a minor component of the mechanism, as the yield of
Mn2(CO)10 is less than 10%.
CONCLUSIONS
■
We have demonstrated that complex 1 is capable of mediating a
water-splitting reaction and that the yields of H2 and H2O2 with
1 are the highest compared to other monomeric L2XMn(CO)3
complexes, possibly because 1 contains water-derived ligands
prior to photolysis. After photolysis of a CO from 1, we
propose that 1-CO undergoes disproportionation to yield Mn0
and 3. The former is responsible for the observed Mn2(CO)10
in post-irradiated solutions of 1. One H2 production pathway
does not require water that we hypothesize originates from H·
release from 3 to form 2 and one-half H2. Labeling studies
confirm that H2 is derived from 1, but the distribution of
dihydrogen isotopes obtained when d4-1 is used suggests that 1
can exchange protons with water and possibly toluene, and we
are currently investigating the scrambling chemistry. The Mn0
species undergo a separate set of photochemical or thermal
reactions with H2Ox (x = 1,2) that also form H2. The lack of
incorporation of label into H2O2 when 18O-1 is used
While the exact nature of the O−O bond-forming step is not
certain, we cast doubt on the formation of free hydroxide
radicals, because hydroxylated organics and {DMPO−OH}· are
not observed.43,34 Both 1 and water are necessary for H2O2
formation. Hence, species uniquely derived from complex 1 are
I
DOI: 10.1021/acs.inorgchem.7b01483
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
demonstrates that the water-oxidation half-reaction occurs from
a species derived from 1-CO (or 3) reacting with water.
Therefore, it may not be necessary to install water-derived
ligands prior to photon absorption for the water-oxidation step.
Specific details surrounding the O−O bond-forming step
remains elusive, but we cast doubt on the viability of pathways
involving free hydroxide radicals by the lack of {DMPO−OH}·
and hydroxylated organic products.
REFERENCES
■
(1) Piers, W. E. Future Trends in Organometallic Chemistry:
Organometallic Approaches to Water Splitting. Organometallics 2011,
30, 13−16.
(2) Klahn, M.; Beweries, T. Organometallic water splitting−from
coordination chemistry to catalysis. Rev. Inorg. Chem. 2014, 34, 177−
198.
(3) Kohl, S. W.; Weiner, L.; Schwartsburd, L.; Konstantinovski, L.;
Shimon, L. J. W.; Ben-David, Y.; Iron, M. A.; Milstein, D. Consecutive
Thermal H2 and Light-Induced O2 Evolution from Water Promoted
by a Metal Complex. Science 2009, 324, 74−77.
(4) Yang, X.; Hall, M. B. Mechanism of Water Splitting and Oxygen−
Oxygen Bond Formation by a Mononuclear Ruthenium Complex. J.
Am. Chem. Soc. 2010, 132, 120−130.
(5) Kunkely, H.; Vogler, A. Water Splitting by Light with Osmocene
as Photocatalyst. Angew. Chem., Int. Ed. 2009, 48, 1685−1687.
(6) Chen, Y.; Han, J.; Fang, W.-H. Mechanism of Water Oxidation to
Molecular Oxygen with Osmocene as Photocatalyst: A Theoretical
Study. Inorg. Chem. 2012, 51, 4938−4946.
(7) Kessler, M.; Hansen, S.; Hollmann, D.; Klahn, M.; Beweries, T.;
Spannenberg, A.; Bruckner, A.; Rosenthal, U. Synthesis of Cp* Ti-
(OTf) and Its Reaction with Water. Eur. J. Inorg. Chem. 2011, 2011,
627−631.
(8) Godemann, C.; Hollmann, D.; Kessler, M.; Jiao, H.;
Together with the results from photochemical water-splitting
with L2XMn(CO)3 complexes, the mechanism we propose
suggests the necessary formation of unsaturated {L2Mn(CO)2-
hydroxido}n species to accommodate additional water ligands
for the water-oxidation reaction. The relatively high yield of
H2O2 from species that contain Brønsted-Lowry basic ligands,
such as cyclopentadienide, compared to nonbasic ligands
supports this premise. Complex 1 already contains hydroxido
moieties and hence is uniquely suited for the formation of these
L2Mn(CO)2-hydroxido units and produces the most H2O2.
Additionally, disproportionation of photolyzed MnI compounds
is a recurring outcome for a variety of L2XMn(CO)3 complexes,
including 1, and appears to be important for the water-
reduction half reaction. Thus, future investigations will focus on
developing ligands that facilitate or inhibit disproportionation
and also proton transfer as to arrive at metastable L2Mn-
(CO)2−hydroxido complexes. Finally, the H2O2 product in the
photochemical reaction of 1 is destroyed by the Mn0 species
that result from the disproportionation indicating the need for a
more robust system or organometallic catalase-like molecules to
achieve turnover conditions, and we are actively pursuing these
strategies in our laboratory.
̈
2
Spannenberg, A.; Bruckner, A.; Beweries, T. A Model of a Closed
Cycle of Water Splitting Using ansa-Titanocene(III/IV) Triflate
Complexes. J. Am. Chem. Soc. 2015, 137, 16187−16195.
̈
(9) Kee, J. W.; Tan, Y. Y.; Swennenhuis, B. H. G.; Bengali, A. A.; Fan,
W. Y. Hydrogen Generation from Water upon CpMn(CO)3
Irradiation in a Hexane/Water Biphasic System. Organometallics
2011, 30, 2154−2159.
(10) Esswein, A. J.; Nocera, G. D. Hydrogen Production by
Molecular Photocatalysis. Chem. Rev. 2007, 107, 4022−4047.
(11) Selected reviews: (a) Blakemore, J. D.; Crabtree, R. H.; Brudvig,
G. W. Molecular Catalysts for Water Oxidation. Chem. Rev. 2015, 115,
ASSOCIATED CONTENT
■
12974−13005. (b) Karkas, M. D.; Verho, O.; Johnston, E. V.;
̈
̈
S
* Supporting Information
Åkermark, B. Artificial Photosynthesis: Molecular Systems for
Catalytic Water Oxidation. Chem. Rev. 2014, 114, 11863−12001.
(c) Najafpour, M. M.; Renger, G.; Holynska, M.; Moghaddam, A. N.;
Aro, E.-M.; Carpentier, R.; Nishihara, H.; Eaton-Rye, J. J.; Shen, J.-R.;
Allakhverdiev, S. I. Manganese Compounds as Water-Oxidizing
Catalysts: From the Natural Water-Oxidizing Complex to Nanosized
Manganese Oxide Structures. Chem. Rev. 2016, 116, 2886−2936.
(12) (a) Tan, K. Y. D.; Kee, J. W.; Fan, W. Y. CpMn(CO)3-Catalyzed
Photoconversion of Thiols into Disulfides and Dihydrogen. Organo-
metallics 2010, 29, 4459−4463. (b) Tan, K. Y. D.; Teng, G. F.; Fan, W.
Y. Photocatalytic Transformation of Organic and Water-Soluble Thiols
into Disulfides and Hydrogen under Aerobic Conditions Using
Mn(CO)5Br. Organometallics 2011, 30, 4136−4143.
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.inorg-
chem.7b01483.
Spectra obtained by NMR, UV−vis, IR, and EPR;
tabulated data including results from photochemical
experiments, yields produced from UV−vis irradiations,
comparison of selected experimental bond lengths, EPR
parameters from literature and simulation, and calculated
excited states; additional references, input file for
TDDFT calculations, qualitative calculated frontier
molecular orbital splitting diagram (PDF)
(13) Zhang, Z.; Li, W.; Liu, J.; Chen, X.; Bu, Y. DFT studies on the
mechanism of the conversion of thiols into disulfides and dihydrogen
catalyzed by CpMn(CO)3 complex. J. Organomet. Chem. 2012, 706−
707, 89−98.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: DCLacy@Buffalo.edu.
ORCID
Karthika J. Kadassery: 0000-0003-2065-1356
Anthony F. Cannella: 0000-0001-6519-2443
David C. Lacy: 0000-0001-5546-5081
(14) Examples of X2 elimination: (a) Cook, T. R.; Surendranath, Y.;
Nocera, D. G. Chlorine Photoelimination from a Diplatinum Core:
Circumventing the Back Reaction. J. Am. Chem. Soc. 2009, 131, 28−29.
(b) Teets, T. S.; Nocera, D. G. Halogen Photoreductive Elimination
from Gold(III) Centers. J. Am. Chem. Soc. 2009, 131, 7411−7420.
(c) Powers, D. C.; Hwang, S. J.; Anderson, B. L.; Yang, H.; Zheng, S.-
L.; Chen, Y.-S.; Cook, T. R.; Gabbaï, F. P.; Nocera, D. G.
Stereoelectronic Effects in Cl2 Elimination from Binuclear Pt(III)
Complexes. Inorg. Chem. 2016, 55, 11815−11820. Examples of H2
elimination from similar platforms: (d) Heyduk, A. F.; Nocera, D. G.
Hydrogen Produced from Hydrohalic Acid Solutions by a Two-
Electron Mixed-Valence Photocatalyst. Science 2001, 293, 1639−1641.
(15) Geoffroy, G. L.; Wrighton, M. S. Organometallic Photochemistry;
Academic Press: New York, 1979.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
D.C.L thanks the Univ. at Buffalo and the State Univ. of New
York Research Foundation for support and the Univ. at Buffalo
Center for Computational Research for access to computation
facilities.
(16) Kadassery, K. J.; Dey, S. K.; Friedman, A. E.; Lacy, D. C.
Exploring the Role of Carbonate in the Formation of an Organo-
J
DOI: 10.1021/acs.inorgchem.7b01483
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
manganese Tetramer. Inorg. Chem. DOI: 201710.1021/acs.inorg-
chem.7b01438.
Using Metallosurfactant Catalysts and Oxalic Acid. Inorg. Chem. 2017,
DOI: 10.1021/acs.inorgchem.7b00463.
(17) For (py)2MnBr(CO)3 synthesis: Pons, M.; Herberich, G. E.
Inorg. Synth. 2014, 36, 148−149. For synthesis of (Tp)Mn(CO)3
(trispyrazolyl borate = Tp), CptetMn(CO)3, and Cp*Mn(CO)3: Kirk,
P.; Castellani, M. P. Inorg. Synth. 2014, 36, 62−64.
(18) Reger, D. L.; Grattan, T. C.; Brown, K. J.; Little, C. A.; Lamba, J.
J. S.; Rheingold, A. L.; Sommer, R. D. Syntheses of tris(pyrazolyl)-
methane ligands and {[tris(pyrazolyl)methane]Mn(CO)3}SO3CF3
complexes: comparison of ligand donor properties. J. Organomet.
Chem. 2000, 607, 120−128.
(30) Nakamoto, K. Infrared Spectra of Inorganic and Coordination
Compounds, 2nd ed.; Wiley-Interscience: New York, 1970.
(31) THF (60 min) = 35% H2 + 7.1 equiv of CO; MeCN (18 h) =
6% H2 + 16% CH4 (no ethane); toluene (120 min to 18 h, broadband)
= 85% 21 H2 + 9.3 2 equiv of CO; benzene (120 min, broadband
or 300−375 nm) = 52% 12 H2 + 7.7 1 equiv of CO.
(32) Holman, K. T.; Zaworotko, M. J. Crystal and molecular
structure of [Mn(CO)3(μ3-OH)]4. J. Chem. Crystallogr. 1995, 25, 93−
95.
(33) Bamford, C. H.; Coldbeck, M. Evidence for the formation of the
triaquatricarbonylmanganese(I) cation and related derivatives from
pentacarbonylchloromanganese. J. Chem. Soc., Dalton Trans. 1978, 4−
8.
(19) Bourrez, M.; Molton, F.; Chardon-Noblat, S.; Deronzier, A.
[Mn(bipyridyl)(CO)3Br]: An Abundant Metal Carbonyl Complex as
Efficient Electrocatalyst for CO2 Reduction. Angew. Chem., Int. Ed.
2011, 50, 9903−9906.
(34) Grela, M. A.; Coronel, M. E. J.; Colussi, A. J. Quantitative Spin-
Trapping Studies of Weakly Illuminated Titanium Dioxide Sols.
Implications for the Mechanism of Photocatalysis. J. Phys. Chem. 1996,
100, 16940−16946.
(35) Buettner, G. R. Spin Trapping: ESR Parameters of Spin
Adducts. Free Radical Biol. Med. 1987, 3, 259−303.
(20) Kurtz, D. A.; Dhakal, B.; Hulme, R. J.; Nichol, G. S.; Felton, G.
A. N. Correlations between photophysical and electrochemical
properties for a series of new Mn carbonyl complexes containing
substituted phenanthroline ligands. Inorg. Chim. Acta 2015, 427, 22−
26.
(21) Hilliard, C. R.; Bhuvanesh, N.; Gladysz, J. A.; Blumel, J.
̈
(36) When DMPO is not included under the normal set of
conditions studied here, benzyl radicals that form are oxidized and
subsequently produce diaryl methanes via Friedel−Crafts alkylation.
This indicates that small amounts of high-valent (O.S. ≥ 2) manganese
are formed in photolyses of 1: (a) Lockwood, M. A.; Wang, K.; Mayer,
J. M. Oxidation of Toluene by [(phen)2Mn(μ-O)2Mn(phen)2]4+ via
Initial Hydride Abstraction. J. Am. Chem. Soc. 1999, 121, 11894−
11895. (b) Bryant, J. R.; Taves, J. E.; Mayer, J. M. Oxidations of
Hydrocarbons by Manganese(III) Tris(hexafluoroacetylacetonate).
Inorg. Chem. 2002, 41, 2769−2776. (c) Larsen, A. S.; Wang, K.;
Lockwood, M. A.; Rice, G. L.; Won, T.-J.; Lovell, S.; Sadílek, M.;
Synthesis, purification, and characterization of phosphine oxides and
their hydrogen peroxide adducts. Dalton Trans. 2012, 41, 1742−1754.
(22) (a) Monzani, E.; Quinti, L.; Perotti, A.; Casella, L.; Gullotti, M.;
Randaccio, L.; Geremia, S.; Nardin, G.; Faleschini, P.; Tabbì, G.
Tyrosinase Models. Synthesis, Structure, Catechol Oxidase Activity,
and Phenol Monooxygenase Activity of a Dinuclear Copper Complex
Derived from a Triamino Pentabenzimidazole Ligand. Inorg. Chem.
1998, 37, 553−562. (b) Ackermann, J.; Meyer, F.; Kaifer, E.; Pritzkow,
H. Tuning the Activity of Catechol Oxidase Model Complexes by
Geometric Changes of the Dicopper Core. Chem. - Eur. J. 2002, 8, 247.
(23) (a) Ramette, R. W.; Sandford, R. W. Thermodynamics of Iodine
Solubility and Triiodide Ion Formation in Water and in Deuterium
Oxide. J. Am. Chem. Soc. 1965, 87, 5001−5005. (b) Burgess, A. E.;
Davidson, J. C. A Kinetic−Equilibrium Study of a Triiodide
Concentration Maximum Formed by the Persulfate−Iodide Reaction.
J. Chem. Educ. 2012, 89, 814−816. (c) Kim, K.; Yabushita, A.;
Okumura, M.; Saiz-Lopez, A.; Cuevas, C. A.; Blaszczak-Boxe, C. S.;
Min, D. W.; Yoon, H.-I.; Choi, W. Production of Molecular Iodine and
Tri-iodide in the Frozen Solution of Iodide: Implication for Polar
Atmosphere. Environ. Sci. Technol. 2016, 50, 1280−1287.
(24) Creaven, B. S.; Dixon, A. J.; Kelly, J. M.; Long, C.; Poliakoff, M.
Structure and reactivity of (η5-C5H5)Mn(CO)2 in room-temperature
solution. Evidence for formation of a dinuclear intermediate detected
by flash photolysis and time-resolved infrared spectroscopy. Organo-
metallics 1987, 6, 2600−2605.
(25) Stor, G. J.; Morrison, S. L.; Stufkens, D. J.; Oskam, A. The
Remarkable Photochemistry of fac-XMn(CO)3(α-diimine) (X =
Halide): Formation of Mn2(CO)6(α-diimine)2 via the mer Isomer
and Photocatalytic Substitution of X- in the Presence of PR3.
Organometallics 1994, 13, 2641−2650.
(26) Xing, Y. H.; Aoki, K.; Bai, F. Y. Synthesis and Structure of the
Mn(II) Complexes with Tripyrazolylborate Ligands: Mn[HB(pz)3]2
and Mn[HB(3,5-Me2-pz)3]2. Synth. React. Inorg. Met.-Org. Chem. 2004,
34, 1149−1163.
(27) (a) Bolton, J. R.; Stefan, M. I.; Shaw, P.-S.; Lykke, K. R.
Determination of the quantum yields of the potassium ferrioxalate and
potassium iodide−iodate actinometers and a method for the
calibration of radiometer detectors. J. Photochem. Photobiol., A 2011,
222, 166−169. (b) Demas, J. N.; Bowman, W. D.; Zalewski, E. F.;
Velapoldi, R. A. Determination of the quantum yield of the ferrioxalate
actinometer with electrically calibrated radiometers. J. Phys. Chem.
1981, 85, 2766−2771.
̌
Turecek, F.; Mayer, J. M. Hydrocarbon Oxidation by Bis-μ-oxo
Manganese Dimers: Electron Transfer, Hydride Transfer, and
Hydrogen Atom Transfer Mechanisms. J. Am. Chem. Soc. 2002, 124,
10112−10123.
(37) Several methods for the quantification of H2O2 are summarized
in ref 9 and include alkaline potassium hexacyanoferrate, acidified
potassium chromate, acidified potassium iodide, and Fenton reagent
test. In our hands, these methods gave false positives due to
manganese-containing oxidants present in analyte solutions.
(38) (a) Prinz, U.; Koelle, U.; Ulrich, S.; Merbach, A. E. The
Organometallic fac-[(CO)3Mn(H2O)3]+ Aquaion: Base-Hydrolysis
and Kinetics of H2O-Substitution. Inorg. Chem. 2004, 43, 2387−
2391. (b) Grundler, P. V.; Helm, L.; Alberto, R.; Merbach, A. E.
Relevance of the Ligand Exchange Rate and Mechanism of fac-
[(CO)3M(H2O)3]+ (M = Mn, Tc, Re) Complexes for New
Radiopharmaceuticals. Inorg. Chem. 2006, 45, 10378−10390.
(39) Gligorovski, S.; Strekowski, R.; Barbati, S.; Vione, D.
Environmental Implications of Hydroxyl Radicals (·OH). Chem. Rev.
2015, 115, 13051−13092.
(40) McGrady, J. E. Periodic trends in metal-metal bonding in
cubane clusters, (C5H5)4M4E4 [M = Cr, Mo, E = O, S]. J. Chem. Soc.,
Dalton Trans. 1999, 1393−1399.
(41) (a) Hummel, P.; Oxgaard, J.; Goddard, W. A., III; Gray, H. B.
Ligand-Field Excited States of Metal Hexacarbonyls. Inorg. Chem.
2005, 44, 2454−2458. (b) Hartwig, J. F. Organotransition Metal
Chemistry, From Bonding to Catalysis; University Science Books: Mill
Valley, CA, 2010.
(42) (a) Yardley, J. T.; Gitlin, B.; Nathanson, G.; Rosan, A. M.
Fragmentation and molecular dynamics in the laser photodissociation
of iron pentacarbonyl. J. Chem. Phys. 1981, 74, 370−378.
(b) Leadbeater, N. Enlightening organometallic chemistry: the
photochemistry of Fe(CO)5 and the reaction chemistry of unsaturated
iron carbonyl fragments. Coord. Chem. Rev. 1999, 188, 35−70.
(43) Hatipoglu, A.; Vione, D.; Yalci̧ n, Y.; Minero, C.; Çinar, Z.
Photo-oxidative degradation of toluene in aqueous media by hydroxyl
radicals. J. Photochem. Photobiol., A 2010, 215, 59−68.
(28) Kee, J. W.; Chong, C. C.; Toh, C. K.; Chong, Y. Y.; Fan, W. Y.
Stoichiometric H2 production from H2O upon Mn2(CO)10photolysis.
J. Organomet. Chem. 2013, 724, 1−6.
(29) Kagalwala, H. N.; Chirdon, D. N.; Mills, I. N.; Budwal, N.;
Bernhard, S. Light-Driven Hydrogen Generation from Microemulsions
K
DOI: 10.1021/acs.inorgchem.7b01483
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
(44) Jin, Z.; Wen, Y.; Xiong, L.; Yang, T.; Zhao, P.; Tan, L.; Wang,
T.; Qian, Z.; Su, B.-L.; He, Q. Intratumoral H2O2-triggered release of
CO from a metal carbonyl-based nanomedicine for efficient CO
therapy. Chem. Commun. 2017, 53, 5557−5560.
L
DOI: 10.1021/acs.inorgchem.7b01483
Inorg. Chem. XXXX, XXX, XXX−XXX