Mechanistically Driven Control over Cubane Oxo Cluster Catalysts

Article abstract of DOI:10.1021/jacs.9b01356

Predictive and mechanistically driven access to polynuclear oxo clusters and related materials remains a grand challenge of inorganic chemistry. We here introduce a novel strategy for synthetic control over highly sought-after transition metal {M₄

Full text of DOI:10.1021/jacs.9b01356

Article
pubs.acs.org/JACS
Cite This: J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Mechanistically Driven Control over Cubane Oxo Cluster Catalysts
Fangyuan Song,† Karrar Al-Ameed,†,‡ Mauro Schilling,† Thomas Fox,† Sandra Luber,†
and Greta R. Patzke*,†
†Department of Chemistry, University of Zurich, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland
‡Faculty of Science, University of Kufa, 54001 Najaf, Iraq
S
* Supporting Information
ABSTRACT: Predictive and mechanistically driven access to poly-
nuclear oxo clusters and related materials remains a grand challenge of
inorganic chemistry. We here introduce a novel strategy for synthetic
control over highly sought-after transition metal {M4O4} cubanes. They
attract interest as molecular water oxidation catalysts that combine
features of both heterogeneous oxide catalysts and nature’s cuboidal
{CaMn4O5} center of photosystem II. For the first time, we
demonstrate the outstanding structure-directing effect of straightfor-
ward inorganic counteranions in solution on the self-assembly of oxo
clusters. We introduce a selective counteranion toolbox for the
controlled assembly of di(2-pyridyl) ketone (dpk) with M(OAc)2 (M = Co, Ni) precursors into different cubane types.
Perchlorate anions provide selective access to type 2 cubanes with the characteristic {H2O-M2(OR)2-OH2} edge-site, such as
[Co4(dpy-C{OH}O)4(OAc)2(H2O)2](ClO4)2. Type 1 cubanes with separated polar faces [Co4(dpy-C{OH}O)4(L2)4]n+ (L2
= OAc, Cl, or OAc and H2O) can be tuned with a wide range of other counteranions. The combination of these counteranion
sets with Ni(OAc)2 as precursor selectively produces type 2 Co/Ni-mixed or {Ni4O4} cubanes. Systematic mechanistic
experiments in combination with computational studies provide strong evidence for type 2 cubane formation through reaction
of the key dimeric building block [M2(dpy-C{OH}O)2(H2O)4]2+ with monomers, such as [Co(dpy-C{OH}O)(OAc)(H2O)3].
Furthermore, both experiments and DFT calculations support an energetically favorable type 1 cubane formation pathway via
direct head-to-head combination of two [Co2(dpy-C{OH}O)2(OAc)2(H2O)2] dimers. Finally, the visible-light-driven water
oxidation activity of type 1 and 2 cubanes with tuned ligand environments was assessed. We pave the way to efficient design
concepts in coordination chemistry through ionic control over cluster assembly pathways. Our comprehensive strategy
demonstrates how retrosynthetic analyses can be implemented with readily available assembly directing counteranions to
provide rapid access to tuned molecular materials.
interdisciplinary interest, e.g., as catalytic model systems for
nature’s photosystem II23 or as magnetic materials.24,25 We
introduce a toolbox of straightforward inorganic counteranions
as one-pot and highly efficient structure-directing agents for
cubane assembly in aqueous solution. Furthermore, we
investigate the underlying formation mechanisms that provide
controlled access to a wide spectrum of {M4O4} (M = Co, Ni,
or Co/Ni) cubanes.
Currently, the gap between the growing number of self-
assembled metal-oxo clusters on the one hand and the lack of
truly predictive design concepts is still widening. Intense
investigations into their formation processes are thus required
to develop predictive access strategies. Recent progress
includes detailed speciation, analytical growth tracking as
well as crystallization studies of smaller18 and larger26
polyoxometalates. More insight into the influence of self-
assembly27 on oxo cluster formation compared to external
INTRODUCTION
■
The present global health energy and challenges trigger an
increasing demand for new functional inorganic compounds
and catalysts. In sharp contrast, their formation mechanisms
and truly predictive syntheses remain grand challenges of
inorganic chemistry. Considerable progress has been reported
in computer-aided solid state and organic synthesis,1−3 and
major progress in synthetic coordination chemistry is expected
from forthcoming machine learning approaches.
Polynuclear transition metal-oxo complexes and polyoxo-
metalates are prominent targets in current catalytic, bio-
inorganic, medicinal, and materials research.4−6 Their wide
application range encompasses, for example, catalytic water
splitting,7−12 new antimicrobial agents,13 and single molecule
magnets.14 However, they also embody the present challenges
of predictive coordination chemistry, because the formation
pathways of most polynuclear transition metal oxo clusters
remain widely unknown. This makes their full exploration and
systematic access to all their attractive properties difficult.15−22
Here we present an unprecedented level of synthetic control
over transition metal cubane oxo clusters that keep attracting
Received: February 4, 2019
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.9b01356
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
chemical stimuli is now indispensable for convenient toolbox
protocols.
RESULTS AND DISCUSSION
■
Cubane Structures: Type 1 and Type 2. Structures. All
The considerable templating and recognition properties of
inorganic counteranions emerged from supramolecular chem-
istry research28−34 and attracts increasing attention in bio-,
environmental, and inorganic chemistry.31,35−39 However, the
full structure-directing potential of counteranions is still far
from explored. The majority of synthetic studies are focused on
solid state anion template effects on the crystal structure of
(oxo) cluster materials,40 e.g., halide-templated assemblies of
first row transition metal ions.41 Far less is known about
anionic structure-directing control over oxo clusters in situ, i.e.,
during their assembly processes in solution.
cubanes discussed in this study belong to two different types (1
and 2) with alternative coordination modes of the primary
dpy-C{OH}O− ligand to the {M4O4} core. The general type 1
and type 2 structures and the different accessibilities of their
polar faces are illustrated in Figure 1 (CIF files for all cubanes
are provided in the SI).
Understanding the influence of anions on cluster formation
pathways with full computational analyses first requires far
more fundamental insight into the basic physicochemical
properties of hydrated anions, beyond the present state. In
sharp contrast to the well studied hydration behavior of
cations, such experimental studies on anions in aqueous
solutions still remain quite limited due to the oftentimes very
fast exchange rates (ps regime) of their weakly bound water
ligands.42 Structural information on anion environments in
solution can be obtained from large angle X-ray or neutron
scattering, respectively.43 The combination of such advanced
analyses with simulations of hydrated anionic structures in
solution and their water exchange mechanisms44 currently
paves the way to their more comprehensive understanding.
Recent illustrative examples include modeling studies on four
45
45
−
−
counteranions used in the present work (ClO4 , ClO3 ,
46
BF4 , and hexafluorophosphate46).
−
Transition metal and heterometallic47−51 cubanes are
attractive and sought-after targets for new design approaches
in coordination chemistry. They are quite versatile catalytic
model systemsboth for nature’s cuboidal and heterometallic
{CaMn4O5} oxygen evolving complex (OEC) of photosystem
II52−55 and for crucial surface motifs of oxide-based water
oxidation catalysts (WOCs).56−58 While much emphasis has
been placed on elucidating the oxygen evolution pathways of
synthetic cubane clusters, considerably less is known about
their own formation processes.59−63 Heterometallic cubanes
provide particularly excellent opportunities to study synergistic
effects in oxide-related as well as in bioinspired molecular
catalysts,64 but their targeted construction remains demand-
ing.65 To the best of our knowledge, the influence of inorganic
anions on the assembly pathways of oxo clusters in solution
remains rather unexplored. We here demonstrate the potential
of straightforward inorganic anions for predictive access to
cubanes, providing inspirational input for forthcoming machine
learning approaches.
To this end, model systems with Co(II)- and Ni(II)-salt
precursors and di(2-pyridyl) ketone (dpk) as ligand for cubane
formation were operated with systematically varied inorganic
counteranion types and concentrations. While dpk has been
identified as an efficient ligand for transition metal cubane
formation in previous studies,66−69 the overarching, mechanis-
tically driven design of transition metal oxo cluster remains
highly sought-after. We first provide a comprehensive roadmap
to different types of {M4O4} cubane oxo clusters and their
heterometallic analogues from M(OAc)2/dpk/NaX (M = Co,
Ni; X = ClO4, ClO3, BF4, PF6, NO3, and Cl) model systems.
Furthermore, we proceed to their mechanistic and computa-
tional analyses and conclude with key visible-light-driven water
oxidation performance data.
Figure 1. Structural features of type 1 (a) and 2 (b) cubanes (ligand
hydrogen atoms and counteranions are omitted for clarity; Co: dark
blue, C: gray, O: red, N: green; inset in (a) = structure of the
hydrolyzed dpk ligand).
Nomenclature. For type 1 cubanes, type and number of
secondary ligands (OAc− and Cl−) and of their corresponding
isolated counteranions are successively abbreviated as in the
following example: 1(type)-3OAc(three secondary acetate
−
ligands)-BF4(one BF4 counteranion), as in [Co4(dpy-C{OH}-
O)4(OAc)3(H2O)]BF4 (cf. also Figure 3c).
For mixed Co/Ni type 1 cubanes, their general or
specifically determined compositions are indicated likewise,
such as in 1-CoxNi4−x-3OAc-NO3 representing the mixed
cubane [CoxNi4−x(dpy-C{OH}O)4(OAc)3(H2O)]NO3.
In type 2 cubane compounds, “μ-OAc” after the initial 2
refers to the presence of a bridging acetate ligand, followed by
“CoxNi4−x” or “Ni4” and the respective counteranion types,
e.g., 2-μ-OAc-CoxNi4−x-ClO4 abbreviating [CoxNi4−x(dpy-
C{OH}O)4(OAc)2(μ-OAc)]ClO4 (Figure 2b, right; μ = μ2
in the following).
Finally, short names of cubanes with the characteristic
{H2O-Co2(OR)2-OH2} edge-site motif are given without
B
DOI: 10.1021/jacs.9b01356
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 3. General {Co4O4} cubane syntheses (a) in the absence of
additional counteranions, (b) with Cl−, (c) in the presence of X− (X−
−
−
−
−
−
= ClO3 , NO3 , BF4 , and PF6 ), and (d) with ClO4 .
It is completed with the formal addition of an aqua ligand to
both M1 and M2 and rotation of the remaining two acetate
groups to an approximately perpendicular position facing the
{M1O1M2O2} plane (Figure 2a).
Figure 2. (a) Formal structural transformation from [M4(dpy-
C{OH}O)4 (OAc)4 ] (type 1) to [M4 (dpy-C{OH}-
O)4(OAc)2(H2O)2]2+ (type 2). (b) Structural transition from cubane
cations [M4(dpy-C{OH}O)4(OAc)2(H2O)2]2+ (2-edge site) to
[M4(dpy-C{OH}O)4(OAc)2(μ-OAc)]+ (2-μ-OAc).
2-μ-OAc Cubanes. Furthermore, type 2 cubanes can be
selectively transferred with specific counteranion concentra-
tions (cf. Scheme 1) into their 2-μ-OAc analogues with
bridging acetate groups. This transformation involves replace-
ment of their aqua ligands on the {H2O-Co2(OR)2-OH2}
edge-site with bidentate acetate ligands, while slightly turning
the monodentate acetate ligands toward the adjacent -OH on
the same polar faces (Figure 2b). The resulting different
position of the monodentate acetate ligand in 2-μ-OAc-
CoxNi4−x (Figures 2b and S7) indicates some positional
freedom in the type 2 cubanes. This also gives rise to the two-
component positional disorder of the monodentate acetate
moieties in the 2-CoxNi4−x-ClO4, -ClO3, and -BF4 series (see
the corresponding CIF file in the SI for details).
Crystal Packing and Role of the Counteranions. Detailed
discussions of anionic disorder, extended hydrogen bonding,
and chirality within type 1 and type 2 cubanes are given in the
SI (cf. p. S28). Most importantly, the anionic positions in their
solid state structures do not display substantial differences
between the individual cubanes.
explicitly mentioning their “μ-OAc” ligands, e.g., 2-CoxNi4−x
-
BF4 for [CoxNi4−x(dpy-C{OH}O)4(OAc)2(H2O)2](BF4)2
(Figure 2b, left).
Structural Characterization of Type 1 Cubanes. All
type 1 cubanes in the present study were structurally
characterized by means of single-crystal X-ray diffraction as
well as with select bulk analytical techniques (Tables S1, S2,
S9, and Scheme S1). Their cobalt centers are coordinated to
three oxygen atoms and two nitrogen atoms of three and two
dpy-C{OH}O− ligands, respectively; in a η:1η3:η1 fashion
(Figure S1). Together with a secondary ligand L2, the four
octahedrally coordinated cobalt centers form [Co4(dpy-
C{OH}O)4(L2)4]n+ type 1 cubanes with two separated polar
faces (Figure 1a). Each of them is equipped with two
secondary L2 ligands (L2 = monodentate OAc−, monodentate
OAc− and H2O, and Cl− for 1-4OAc, 1-2OAc and 1-3OAc,
and 1-4Cl, respectively). The L2 types and ratios can be
controlled through selecting appropriate anion types and
concentrations during synthesis (cf. Scheme 1, Figure S2 and
section below). Systematic replacement of the four acetate
ligands of 1-4OAc with one or two aqua- or four chloride
ligands thereby affords the 1-3OAc, 1-2OAc, and 1-4Cl
cubanes (Figure 3).
Structural Characterization of Type 2 Cubanes. Type
2 cubanes were structurally characterized with single-crystal X-
ray diffraction throughout, in combination with appropriate
bulk analytical techniques (Tables S3−S7, S10−14, and
Scheme S1).
2-Edge-Site Cubanes. The upper polar {M-O-M-O} face in
the type 2 cubanes (cf. Figure 1b) bears the secondary ligands
(referred to as edge-site in the following), while the lower
plane is solely coordinated to dpy-C{OH}O− ligands (Figure
1b). The formal transition from type 1 to type 2 cubanes thus
involves the dissociation of two acetate groups (on the same
polar side) while recoordinating N1 and N4 to M3 and M4,
respectively.
Therefore, the selective access to tuned cubane structures as
outlined in the following section (cf. Scheme 1) cannot be
ascribed to anionic templating in the solid state.
Predictive Anion-Controlled Access to Type 1 and
Type 2 Cubanes. General Synthesis and Characterization.
We have developed a predictive, counteranion-directed access
to type 1 and type 2 cubanes from the convenient, one-pot
reaction of M(OAc)2 as M2+ source and dpk as in aqueous
solution at room temperature. Scheme 1 illustrates how the
experimental protocols for types 1 and 2 given in the SI (p.
S4−S6) can be adjusted with select counteranion concen-
trations to afford a wide range of cubane targets. The bulk
purity of all cubane types obtained from the strategy shown in
Scheme 1 was corroborated with four different analytical
protocols based on elemental analyses, powder X-ray
1
diffraction (PXRD), H NMR and Raman spectroscopy. The
choice of these methods is explained in detail in the SI (p.
S34−S35, where Scheme S1 specifies the characterization
techniques applied to each cubane type). Additionally, further
checks of cubane stabilities and Co/Ni-substitution patterns
were performed with HR-ESI-MS (cf. p. S30−S34 and
C
DOI: 10.1021/jacs.9b01356
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 1. Synthetic Fields for Counteranion-Directed
Access to Type 1/2 {Co4O4}, {CoxNi4−xO4}, and Type 2
a
{Ni4O4} Cubanes
Figure 4. 1H NMR spectrum of 1-2OAc-2BF4 and peak assignments
on the basis of T1.
crystallographic and analytical data). Consequently, intermedi-
−
ate counteranion amounts of 2.5−7.5 equiv BF4 give rise to
coexistence of both compounds, as is evident from the
1
presence of both sets of proton signals in the respective H
NMR spectra (cf. Figure S27 and Table S15 for the peak
assignments). This coexistence of 1-2OAc-2BF4 and 1-3OAc-
−
BF4 as products of the intermediate 2.5−7.5 equiv BF4 route
1
is evident from the virtually identical H NMR spectrum of a
mixture of pure 1-2OAc-2BF4 and 1-3OAc-BF4 (Figure S28
and Table S15) compared to that of the as-synthesized mixed
product in Figure S27 with corresponding signal intensities.
Anion Selectivity. A related structure-directing effect on the
number of associated acetate ligands was observed for
a
All M(OAc)2 and NaX amounts are given in mmol, referring to the
−
−
−
standard experimental procedure provided in the SI. Vertical yellow
lines indicate defined NaX concentration thresholds for the respective
products (for an overview of analytical characterizations of the
products, cf. Scheme S1 in the SI).
systematically varied amounts of NO3 , ClO3 , and PF6
(Scheme 1, Figures S29−S31 and Table S15). In contrast to
1-2OAc-2BF4, however, manifold efforts to obtain phase pure
1-2OAc-2NO3, 1-2OAc-2ClO3, and 1-2OAc-2PF6 remained
unsuccessful so far, as well as attempts to directly access pure
1-2OAc from saturated sodium salt solutions of these anions.
S60−S62 in the SI for a complete data evaluation). The level
of crystallographic and structural novelty is specified in Table
S19 for each of the newly reported cubane structures.
For general synthetic operations and analytical information,
we refer to these comprehensive resources in the SI. In the
following, we selectively illustrate examples of counteranion
control over the different cubane types, which reveal
underlying parameters of our new predictive strategy.
−
When applying ClO4 , a structural transformation from type 1
to 2 set in for >0.5 equiv NaClO4 (Scheme 1, cf. Figures S32,
S33, and Table S16 for the component analysis), and pure 2-
Co4-ClO4 is accessible with ≥1.0 equiv NaClO4 (Figure
S45a).56
The exclusive formation of 1-4OAc from synthetic protocols
involving no such counteranions is a strong indication that
they indeed control the number of acetate ligands in the type 1
cubanes through concentration-dependent dissociation pro-
cesses (Figure 3a, cf. Figure S3a and Table S1 for
crystallographic data and Figure S25 for analytical purity).
Furthermore, Cl− exerts a dual function as both counter-
anion and ligand that directly coordinates to the cobalt centers.
This further extends the above range of type 1 cubanes with a
new, neutral member, namely, [Co4(dpy-C{OH}O)4Cl4], 1-
4Cl (cf. Figures 3b, S3b, and Table S1 for crystallographic
data, and Figures 5 and S52a for analytical characterizations).
Access to Heterometallic Co/Ni Cubanes. Scheme 1
furthermore demonstrates how select mixed Co/Ni cubanes
can be directly accessed from according Co(OAc)2/Ni(OAc)2
precursor mixtures and adjusted concentrations of Ni2+ and
counteranions. All mixed cubanes were checked for bulk purity
with the strategies mentioned in the beginning of this section
(the detailed techniques are given in Scheme S1).
Anionic-Directed Access to {Co4O4} Cubanes. The
formation of phase pure edge-site 2-Co4-ClO4 in the presence
of ≥1.0 equiv ClO4 with all initial M2+ amounts (cf. also
−
below) served as our starting point56 to unravel the
unprecedented control effect of an entire series of counter-
−
−
−
−
anions (ClO3 , NO3 , BF4 , PF6 , and Cl−) on type 1
{Co4O4} cubanes (Figure 3d and Scheme 1).
Ligand Tuning of {Co4O4} Cubanes. As a first representa-
tive example for the manifold reactions summarized in Scheme
1, we demonstrate how the number of coordinated acetate
ligands can be fine-tuned between 1-3OAc and 1-2OAc
cubanes with counteranionic control (Scheme 1, upper
section).
−
In the presence of BF4 , phase pure 1-3OAc-BF4 is
selectively accessible through addition of 1.0 equiv NaBF4
(Scheme 1 and Figure 3c, cf. Figures S26 and S43a for
analytical characterizations). However, a notable increase of
the BF4 amount to 8.75 equiv reduces the acetate content in
the cubane and affords the structural analogue 1-2OAc-2BF4
−
The formation of genuine mixed type 1 Co/Ni cubane cores
was confirmed with HR-ESI-MS analyses of phase pure, single
crystalline samples through applying our previous strategies for
(Scheme 1, Figures 4, S4 and S44b, and Table S1 for
D
DOI: 10.1021/jacs.9b01356
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
anions (Cl− and NO3 ) enable the transformation of 2-μ-OAc-
CoxNi4−x (Ni(OAc)2:Co(OAc)2 ≤ 1:3) into new mixed type 1
cubanes (Scheme 1 and Figures S2, S57−S62); note that
NaNO3 still gave rise to a mixture of 2-μ-OAc-CoxNi4−x-NO3
and 1-CoxNi4−x-3OAc-NO3 (Scheme 1 and Figure S58).
Phase pure 1-CoxNi4−x-4Cl cubanes can be accessed with
−
higher NaCl amounts (≥6.25 equiv), such as 1-Co3.31Ni0.69
-
4Cl emerging from Co(OAc)2/Ni(OAc)2 = 0.6 mmol/0.2
mmol (ICP elemental analyses, cf. also Scheme 1, Figures 5
and S52b). In contrast, addition of only 0.125−0.5 equiv
NaPF6 to the same precursor mixture immediately provided
phase pure 1-Co3.28Ni0.72-3OAc-PF6 (ICP elemental analyses,
Scheme 1, Figures S44a and S52c).
Given that the counteranion concentration is the salient
difference between all these analogous synthetic pathways, we
note that a remarkable extent of anion-induced structure control
in solution is at work here that differs notably from a mere
cocrystallization effect.
The selective influence of the counteranion type in solution is
furthermore illustrated in Scheme 1 within the given
Ni(OAc)2:Co(OAc)2 ratio ranges. An illustrative example is
the absence of the above-described retransformation from type
2 Co/Ni cubanes to type 1 when applying higher NaClO4,
NaClO3, or NaBF4 concentrations (according analyses of
crystalline samples and analytical protocols are given in the SI,
p. S34−S35, Figures S45 and S53).
Overview. In conclusion, we provide counteranion-directed
control over cubanes on three structural levels: (1) differ-
entiation between types 1/2, (2) access to specific mixed type
1 and 2 Co/Ni cubane compositions, and (3) adjustment of
cubane ligand patterns for both types. This new territory of
highly selective functionalities of straightforward inorganic
counteranions in the assembly process of transition metal oxo
clusters is first covered with mechanistic studies in the
following.
Experimental Investigation of Counteranion-Direc-
ted {Co4O4} Cubane Assembly Mechanisms. Strategy.
Understanding the dynamic hydration behavior of anions in
solution remains a fundamental challenge for current analytical
and computational research in its own right, as outlined in the
Introduction.42−46 The absence of such detailed insight
renders full computational analyses of counteranion-directed
cluster assembly processes impossible to date. Likewise, the
development of methodologies for their direct in situ
monitoring is a research topic on its own.
Therefore, we here designed a set of systematic screening
experiments and computational models that sheds light on the
basic role of the counteranions in differentiating between type
1 and 2 cubanes. We started with three basic working
hypotheses about the selective formation of cubane types. For
the sake of conciseness, we here summarize our main
mechanistically relevant results that support our retrosynthetic
concept. We refer to the extensive SI for analytical data of the
compounds discussed in the following.
Figure 5. Experimental PXRD patterns of 1-Co4-4Cl and 1-
Co3.31Ni0.69-4Cl vs the calculated pattern of 1-Co4-4Cl.
analyzing type 2 mixed cubanes.56 For example, the
appearance of isotope patterns at lower m/z ranges than
those of the pure Co containing fragments [Co4(dpy-
C{OH}O)4Cl2]2+ (m/z 5552+, Figures S16 and S60) and
[Co4(dpy-C{OH}O)4(OAc)2]2+ (m/z 5792+, Figures S15 and
S61) indicates the presence of both Ni and Co cations in the
cubane cores of 1-CoxNi4−x-4Cl and 1-CoxNi4−x-3OAc-NO3,
in line with their calculated isotope patterns (Figures S60 and
S61).
Synthetic Trends for Co/Ni Cubanes. First, Ni2+/Co2+
precursor ratios >1:4 generally afford mixed {M4O4} cubanes,
with a strong preference for type 2 over type 1 (Scheme 1 and
Figure S2). In the following, we outline that not only the
counteranion type but also its concentration exerts a decisive
influence on the cubane type.
Counteranion Concentration and 2-μ-OAc-Co/Ni Cu-
banes. An impressive example for the counteranion influence
is the systematic access to a series of 2-μ-OAc-CoxNi4−x
cubanes with bridging acetate ligands, namely to
[CoxNi4−x(dpy-C{OH}O)4(μ-OAc)(OAc)2]ClO4, -ClO3,
-NO3, -BF4, -PF6, and -Cl with low to moderate counteranion
concentrations (Scheme 1, cf. Figure S7 and Table S4 for
crystallographic data and Figures S50, S51, and S54 for
analytical results). Second, notably increased counteranion
−
−
−
concentrations of ClO4 , BF4 , and ClO3 provided direct
access to the corresponding edge-site {CoxNi4−xO4} cubanes,
most likely through dissociation of the bridging acetate ligand
in 2-μ-OAc-CoxNi4−x-ClO4, -BF4, and -ClO3, respectively
(Scheme 1, cf. Figures S6 and Table S3 for the crystallographic
data and Figures S45, S47, S48, and S53 for analytical data).
This proposed pathway is supported by the spontaneous
formation of 2-CoxNi4−x-PF6 from as-synthesized 2-μ-OAc-
CoxNi4−x-PF6 after 1 month storage in its mother liquid (e.g.,
2-Co1.64Ni2.36-PF6, cf. Figure S6d and Table S3 for crystallo-
graphic data and Figures S45d and S53b for analytical purity).
In contrast, such a dissociation of the bridging acetate group
was not observed for 2-μ-OAc-CoxNi4−x-Cl and -NO3, even
when stored in saturated solutions of NaCl or NaNO3.
1. Hypothesis: Crystal Packing Effects. We first proposed
the presence of both type 1 and 2 cubane cations in solution.
Such mixtures would then be driven toward selective
crystallization of the optimal packing motif by the respective
counteranion. However, the formation of pure 1-4OAc (60%
yield on the basis of dpk, Figure 3a) from the standard
synthetic protocol in the absence of further counteranions does
not support this assumption. Furthermore, the synthetic routes
to {Co4O4} (Scheme 1) and mixed Co/Ni cubanes (cf. Co:Ni
Counteranion Concentration and Type 1 Co/Ni Cubanes.
Interestingly, higher concentrations of these specific counter-
E
DOI: 10.1021/jacs.9b01356
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
ratios of 3:1 in Scheme 1) clearly provide counteranion-
dependent access to both type 1 and 2 cubanes. This
underscores their active structure-directing role in solution that
goes beyond mere crystal packing effects. As mentioned in the
structural discussion above and shown in the SI, most
counteranions occupy generic positions in the analogous
cubane crystal structure types, which does not point to any
selective solid state templating effects (cf. Figures S3−S10).
−
2. Hypothesis: ClO4 and Metal Precursors. Both previous
works70,71 and our own UV−vis spectroscopy studies (Figures
6a and S64) indicate an equilibrium between free Co ions
Figure 7. Retrosynthetic analyses of the coordination reaction (a) and
of type 1 and 2 (b,c) cubanes based on the respective dimeric
building block.
Figure 6. UV−vis spectra of 130 mM Co(OAc)2 (aq.) (a) and
Ni(OAc)2 (aq.) (b) vs other cobalt and nickel salts (aq.). UV−vis
spectra of pure 130 mM Co(OAc)2 (aq.) and after addition of
different amounts of NaClO4 (c) and NaNO3 (d).
upon addition of dpk into Co(OAc)2 (aq.) in any given
synthetic route (Figure 8), which is in line with previous
synthetic studies.72
3.1. Different Pathway Options to Types 1 and 2
Cubanes. In the absence of additional counteranions,
[Co2(dpy-C{OH}O)2(OAc)2] can either combine with each
other or react with [Co(OAc)]+ and dpy-C{OH}O− ligands to
form the fully acetate-substituted 1-4OAc cubane (Figure
8a,b). 1-3OAc, 1-2OAc, or 1-4Cl cubanes can then be formed
[Co(H2O)6]2+ (abbreviated as [Co]2+ in the following) and
[Co(H2O)5(OAc)]+ (short: [Co(OAc)]+) in the dissociation
of the precursor material Co(OAc)2 (aq.). A formal
retrosynthetic analysis of both cubane types shows that 1-
4OAc contains only the [Co(OAc)]+ building block, while
type 2 is formally constituted of both [Co]2+ and [Co(OAc)]+
moieties (Figure 7a). As type 2 {Co4O4} cubanes could
selectively be obtained in the presence of NaClO4 (Scheme 1),
−
−
−
through addition of structure-directing ClO3 , BF4 , PF6 ,
NO3 , and Cl− anions. We obtained an indication for an anion-
−
driven OAc− ligand dissociation from type 1 cubanes through
the observed transition from pure 1-3OAc-BF4 products over a
mixture with their 1-2OAc analogues to pure 1-2OAc-2BF4
upon increased counteranion amounts (cf. Scheme 1, upper
entries). Moreover, we did not observe conversion of 1-2OAc-
−
one might argue that higher ClO4 concentrations could
facilitate the release of [Co]2+ from Co(OAc)2 (aq.) toward
the concentrations required for type 2 cubane assembly
(Figure S63b).
−
However, the UV−vis spectrum of Co(OAc)2 + 3 equiv
NaClO4 (aq.) is almost identical to that of pure Co(OAc)2
solutions and differs considerably from Co(ClO4)2 (aq.)
(Figure 6c). As a consequence, the direct interaction between
ClO4 into its type 2-edge-site analogue with excess ClO4 ,
despite its [Co2(dpy-C{OH}O)2(H2O)2] building block. This
is additional support for independent assembly pathways of
type 1 and 2 cubanes. Figure 8 presents three plausible
pathways that are further discussed in the following.
−
ClO4 and the Co(OAc)2 precursor is unlikely to promote
−
type 2 cubane formation.
3.2. Selectivity of ClO4 for Type 2 Cubanes. As outlined
−
3. Hypothesis: Retrosynthetic Approach. Therefore, we
proposed in our main working hypothesis the assembly of type
1 and 2 cubanes from their retrosynthetic building blocks
[Co2(dpy-C{OH}O)2(OAc)2(H2O)2] and [Co2(dpy-C{OH}-
O)2(H2O)4]2+ via different anion-dependent pathways. Both
building blocks are shown in Figure 7b and c and are
abbreviated as [Co2(dpy-C{OH}O)2(OAc)2] and [Co2(dpy-
C{OH}O)2]2+ in the following. We further postulate that the
intermediate [Co2(dpy-C{OH}O)2(OAc)2] is formed first
for hypothesis 2 above, ClO4 does not significantly interfere
with the [Co(OAc)]+/[Co]2+ equilibrium. Nevertheless, we
−
have strong experimental indications that ClO4 can indeed
selectively induce the dissociation of acetate ligands from
cubanes. For example, only >0.31 equiv ClO4− are sufficient to
convert bridged 2-μ-OAc-CoxNi4−x into its edge-site type 2
analogue, while far higher concentrations (5 and/or 6.25
−
−
equiv) of ClO3 and BF4 are required to induce this
conversion (Scheme 1). Further clear evidence for such
F
DOI: 10.1021/jacs.9b01356
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Consequently, type 1 cubanes are more likely formed via the
pathways a and b shown in Figure 8.
3.3. Evidence for the Building Block [Co2(dpy-C{OH}-
O)2(OAc)2(H2O)2]. To further support these proposed for-
mation pathways, we first attempted to isolate the key dimer
[Co2(dpy-C{OH}O)2(OAc)2(H2O)2] from ethanol while
adding LiOH·H2O, following the reported procedure for its
Zn-analogue [Zn2(dpy-C{OMe}O)2Cl2].73 Although this
route remained unsuccessful, we nevertheless obtained a
defect cubane from asymmetric assembly of two dimeric
[Co2(dpy-C{OH}O)2(OAc)2] moieties. Interestingly, this
defect cubane was converted into 1-4OAc after 2 weeks of
storage in its mother liquid (Figure S65, cf. Table S8 for
crystallographic data and Figure S25 for analytical character-
ization). This result supports the crucial role of the building
block [Co2(dpy-C{OH}O)2(OAc)2(H2O)2] in cubane assem-
bly.
One reason for the difficulties in isolating this key building
block may arise from the high reactivity of its η2-oxygen center
toward [Co(OAc)]+ or toward the Co site of another
[Co2(dpy-C{OH}O)2(OAc)2(H2O)2] moiety. To check this
assumption, we synthesized the phenyl-modified dpk ligand
1,1-bis(2-pyridyl)phenylmethanol where the bulky substituent
might prevent the dimerization of the intermediates. This
precursor indeed afforded the expected dimer [Co2(dpy-
C{Ph}O)2(OAc)2] as crucial support for the proposed
pathways via dimeric Co species (Figure S11b and Table S8).
3.4. Key Experimental Evidence for Type 1 and 2
Formation Pathways. Next, we used Co(NO3)2 as acetate-
free Co2+ source to further differentiate between pathways a−c
through excluding any acetate coordination to the dimeric
intermediate [Co2(dpy-C{OH}O)2(H2O)4]2+.
As another strong indication for the assembly of type 2
cubanes via pathway c, the new type 2 cubane [Co4(dpy-
C{OH}O)4(H2O)4](NO3)4 (2-(gem-aqua)-Co4-NO3) with a
gem-aqua {Co2(H2O)4} edge-site was formed as expected
(Figures 10a, cf. Figure S8a and Table S6 for crystallographic
data).
Figure 8. Possible assembly pathways of type 1 (a,b) and type 2
cubanes (c). Aqua ligands on the dimeric intermediates are omitted
for clarity.
−
ClO4 induced acetate ligand dissociation is the facile
transformation of 1-3OAc-BF4 into phase pure 1-2OAc-
2ClO4 within a few hours in the presence of a moderate 10-
−
fold molar excess of ClO4 . In drastic contrast, obtaining pure
−
1-2OAc through BF4 addition requires a massive 700-fold
molar excess (cf. SI for experimental details and Figures 9 and
S34 for analytical characterizations).
Surprisingly, the new compound [Co5(dpy-C{OH}-
O)5(dpy-C{O}2](NO3)3 (2-[Co5]-NO3) with a [Co(dpy-
C{OH}O)2] moiety coordinated to the edge-site cocrystallized
with 2-(gem-aqua)-Co4-NO3 (Figure S12 and Table S7). Pure
2-(gem-aqua)-Co4-NO3 can be obtained through applying a
starting ratio of Co(NO3)2/dpk = 1.5 instead of 1 (Figures S45
and S56).
We then analyzed the influence of acetate anions on the
synthetic pathway through varying the initial [Co]2+/[Co-
(OAc)]+ ratios. First, a [Co]2+/[Co(OAc)]+ ratio >1 (i.e.,
≤0.2 mmol NaOAc/0.4 mmol Co(NO3)2 (aq.)) gave rise to 2-
Co4-NO3, indicating that the dpk ligand reacts more readily
with free [Co2+] according to pathway c (Figures 10a, cf.
Figure S10c and Table S7 for crystallographic data and Figures
S35 and S45 for analytical characterizations). Formation of 2-
Co4-NO3 under these conditions also demonstrates that the
Figure 9. PXRD patterns of cubane products supporting the pathway
of 1-2OAc formation via ClO4 induced dissociation of acetate
ligands from 1-3OAc.
−
−
presence of ClO4 is convenient, but not an absolutely
necessary criterion for type 2 cubane formation.
Second, [Co]2+/[Co(OAc)]+ ratios <1 trigger a change to 1-
3OAc-NO3, which strongly suggests that pathways a or b
prevail when [Co(OAc)]+ is present as major species (cf.
Figure S64 for predominance of [Co(OAc)]+ in analogous
conditions, Figure S5b and Table S2 for crystallographic data
and Figures S38 and S43b for analytical characterizations).
These conclusions were backed up with analogous results of
−
With clear indications for this role of ClO4 at hand, we
conclude that it can also facilitate acetate removal from
[Co2(dpy-C{OH}O)2(OAc)2(H2O)2] to form the building
block [Co2(dpy-C{OH}O)2(H2O)4]2+ (Figure 8c). Given that
almost all perchlorate-assisted routes afford type 2 cubanes
(Scheme 1), we conclude that they are plausibly formed via
pathway c of Figure 8.
G
DOI: 10.1021/jacs.9b01356
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
results can presumably be applied to the assembly of {Ni4O4}
and {CoxNi4−xO4} cubanes as well.
First, the optimized structures of four possible dimer types
with acetate-coordinated Co centers, namely, [Co2(dpy-
C{OH}O)2(μ-OAc)2] (D1), [Co2(dpy-C{OH}O)2(OAc)2]
(D2), [Co2(dpy-C{OH}O)2(OAc)4]2−(D3), and [Co2(dpy-
C{OH}O)2(OAc)2(H2O)2] (D4) (Figure 11) were generated.
Figure 11. Co dimeric building blocks subjected to computational
studies.
The relative energy comparison of these most probable dimers
D1-4 revealed that D4 is indeed the thermodynamically most
stable species (Figure 11). This result corresponds well with its
crucial role that emerged from our experimental probing of the
cubane assembly pathways. The computed reaction energy
suggests that 1-4OAc indeed results from direct head-to-head
combination of two D4 dimers via pathway a (Figures 8a and
S68a), where the dpk ligand adopts a tridentate coordination
mode. Furthermore, the reaction energies for various assembly
pathways of 1-4OAc via different [Co(dpy-C{OH}O)-
(OAc)n(H2O)m]2−n monomers were calculated (Figure S69
and Table S20) and found to be thermodynamically less
feasible than pathway a. All in all, DFT calculations fully
support the experimentally postulated formation pathway a of
1-4OAc (Figures 8a and S68a).
On the basis of the above energy comparisons (summarized
in Table S20), the most stable Co monomer was found to be
[Co(dpy-C{OH}O)(OAc)(H2O)3] arising from [Co-
(H2O)5(μ1-OAc)]+ coordinating with one N and O− of dpy-
C{OH}O− in cis- and trans-position to the OAc− ligand (M12,
Figure S69). Assuming that increasing the ionic strength of the
reaction solution, e.g., through perchlorate addition, encour-
ages M12 formation, these monomers can further react with
D4 via acetate ligand dissociation to afford 2-Co4-ClO4
(Figure S68b). This plausible route corresponds well with
the experimentally postulated pathway c toward type 2 cubane
assembly.
Figure 10. Representations of {Co4O4} (a) and {Ni4O4} (b) cubane
syntheses through addition of 0.4 mmol dpk (aq.) into an aqueous
solution of 0.4 mmol M(NO3)2 in the presence of increasing amounts
of NaOAc.
the same synthetic screenings in the presence of chlorate and
tetrafluoroborate counteranions instead of nitrate (cf. Figure
S5, S10 and Tables S2 and S7 for the crystallographic data and
Figures S36, S37, and S45 for analytical characterizations).
Conclusions for {Co4O4} Cubane Formation. Collected
experimental evidence from the first part of our systematic
mechanistic study clearly supports (a) the active role of
counteranions in the cluster assembly process, (b) the key role
of the [Co2(dpy-C{OH}O)2(OAc)2] and [Co2(dpy-C{OH}-
O)2]2+ building blocks, and (c) their participation in pathways
a/b and c giving rise to type 1 and 2 cubanes, respectively.
Computational Studies. Starting from our comprehensive
experimental support for the assembly of type 1 and 2 cubanes
via their respective dimeric moieties [Co2(dpy-C{OH}-
O)2(OAc)2(H2O)2] and [Co2(dpy-C{OH}O)2(H2O)4]2+, we
applied density functional theory (DFT) computations (cf. SI
for details) to further corroborate our retrosynthetic
hypotheses. First, the calculated thermodynamic stabilities of
various monomeric and dimeric building blocks were
compared and then combined with the experimental
observations to identify the most plausible assembly pathways.
Due to the similar coordination behavior of Co2+ and Ni2+ ions
toward the dpk ligand, calculations were representatively
performed for the cobalt compounds in the following. The
Formation Pathways of Ni-Containing Cubanes.
Influence of Ni2+ and OAc− on Type 2 Cubane Assembly.
With this insight on {Co4O4} cubanes at hand, we reasonably
assumed that type 2 {Ni4O4} cubanes assemble through
pathway c. This agrees well with the observed predominance
of [Ni(H2O)6]2+ (abbreviated as [Ni]2+) when comparing the
UV−vis spectrum of Ni(OAc)2 (aq.) to other Ni2+ salts of
H
DOI: 10.1021/jacs.9b01356
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
strong acids (Figure 6b), as well as with previous studies.74
[Ni]2+ and dpk are very likely to form the main intermediate
[Ni2(dpy-C{OH}O)2(H2O)4]2+ with a strong preference for
type 2 cubane formation (Scheme 1). Indeed, reference
reactions in acetate-free media with Ni(NO3)2 as [Ni]2+ source
afforded the expected gem-aqua {Ni4O4} cubane [Ni4(dpy-
C{OH}O)4(H2O)4](NO3)4, 2(gem-aqua)-Ni4-NO3 (Figure
10b, cf. Figure S8b and Table S6 for crystallographic data
and Figures S45 and S56 for analytical characterization).
Moreover, the {Ni4O4} cubane types obtained from
increasing initial NaOAc contents support pathway c.
Interestingly, the addition of 0.25 equiv NaOAc results in a
new type 2 compound, namely, [Ni4(dpy-C{OH}O)4(OAc)-
(H2O)3](NO3)3 (2(half-gem-aqua)-Ni4-NO3), where one
aqua ligand parallel to the {NiONiO} plane is replaced by
OAc− (Figures 10b, cf. S9 and Table S6 for crystallographic
data and Figure S56 for analytical characterizations). This
indicates a different substitutional behavior of the gem-aqua
ligands on the {Ni2(H2O)4} edge-site toward stronger donor
ligands. Finally, this directed ligand assembly of {Ni4O4}
cubanes as a function of increasing initial OAc− content was
supported through the formation of 2-Ni4-NO3 and 2-μ-OAc-
Ni4-NO3 with 0.5 equiv and ≥0.75 equiv, respectively (Figure
10b, cf. Figure S7c, Table S5, and Table S6 for crystallographic
data and Figures S45, S53, and S55 for analytical character-
izations).
experimentally supported pathway c, even in the absence of
further structure-directing perchlorate anions.
Visible-Light-Driven Water Oxidation Studies. We
furthermore carried out visible-light-driven water oxidation
studies to compare the influence of structural features on the
catalytic activity of specific cubane types. To this end, we
selected a set of non-edge-site, {H2O-Co2(OR)2-OH2} edge-
site, gem-aqua {Co(OH2)4} edge-site, and edge-site-blocked
catalysts, namely, 1-3OAc-ClO4, 2-Co4-ClO4, 2(gem-aqua)-
Co4-NO3, and 2-[Co5]-NO3, respectively.
These cubanes were tested at pH 8.5 in 80 mM borate buffer
solution containing 1 mM [Ru(bpy)3]Cl2 as photosensitizer
and 5 mM Na2S2O8 as sacrificial electron acceptor (Figure 12
Figure 12. Clark electrode kinetics of visible-light-driven water
oxidation of the pristine catalysts (solid curves, left) and of their
respective filtered postcatalytic reaction solutions (dashed curves,
right; conditions: 470 nm LED, 1 mM [Ru(bpy)3]Cl2, 5 mM
Na2S2O8, pH 8.5 80 mM borate buffer). The pH value was readjusted
8.5 by adding solid Na2B4O7·10H2O after each reaction.
Interplay of Ni2+ and Co2+ in Cubane Formation.
Furthermore, the key role of free [M]2+ species in type 2
cubane formation via intermediate [Co2(dpy-C{OH}-
O)2(H2O)4]2+ became evident from the strongly selective
formation of mixed type 2 cubanes even in the presence of low
initial Ni contents (0.6 mmol Co(OAc)2 + 0.2 mmol
Ni(OAc)2, Scheme 1). In light of the above, we assume that
the dpk ligand first reacts with the predominant [Ni]2+ species
to the corresponding Ni-rich [M2(dpy-C{OH}O)2(H2O)4]2+
intermediate to provide type 2 {CoxNi4−xO4} products in the
and Table S18 as well as experimental part of the SI). 2-Co4-
ClO4 with the edge-site motif is more active than the non-
edge-site cubane 1-3OAc-ClO4, which relates well with our
previous studies.56 Furthermore, 2(gem-aqua)-Co4-NO3 ex-
hibited the best catalytic performance with a turnover number
(TON) of 20.8, a turnover frequency (TOF) of 0.27 s−1, and
an O2 evolution yield of 83.0%. This activity is rather high
compared to previous works on analogous cubane
WOCs,54,55,75−77 indicating that the aqua ligand-enriched
gem-aqua is likely to promote the oxygen evolution activity.
Along these lines, 2-[Co5]-NO3 showed the lowest activity
with a TON of 11.5, a TOF of 0.10, and an O2 evolution yield
of 46%. This result further indicates that the accessibility of the
edge-site is indeed relevant for the catalytic activities of the
cubane catalyst series. Our systematic synthetic approach now
provides convenient access to such tuned cubanes for
forthcoming detailed structure−activity studies in their own
right.
Cubane catalyst stability was studied with a sequence of
filtration and activity tests of the recycled postcatalytic reaction
solution.56,78 All cubane catalysts retained most of their
catalytic activities during the recycling runs. In contrast, the
reference tests with Co(OAc)2 mimicking the behavior of in
situ formed CoOx particles showed no activity in the recycling
run (Figure 12). This provided strong evidence that all the
cubanes remained in molecular states after the first catalytic
run.
−
−
presence of ClO3 , BF4 , and PF6− via pathway c. Indeed, the
actual ratios of Co:Ni in all obtained 2-CoxNi4−x-ClO3, -BF4,
and -PF6 always fall below the respective initial ratios (Table
S17). The above-mentioned (section 3.2.) directing effect of
−
ClO4 on cubane fragments and acetate anions apparently
releases higher [Co]2+ concentrations for type 2 cubane
formation. This agrees with the observed higher Co content of
as-synthesized 2-CoxNi4−x-ClO4 compared to the 2-CoxNi4−x
-
ClO3, -BF4, and -PF6 products obtained from identical Co:Ni
starting ratios (ICP-MS measurements in Table S17).
Consequently, relatively low amounts of free [M]2+ should
lead to insufficient concentrations of the type 2 building-block
[M(dpy-C{OH}O)2(H2O)4]2+ in the presence of other
−
−
counteranions than ClO4 . Indeed, we observed that BF4
−
and ClO3 addition led to (partial) type 1 cubane formation
under analogous conditions, such as a mixture of 1-CoxNi4−x
-
3OAc-BF4 and 2-μ-OAc-CoxNi4−x-BF4 obtained from 0.7
mmol Co(OAc)2 + 0.1 mmol Ni(OAc)2 with 3 mmol NaBF4
(Figures S66a and S67). Similarly, 1-CoxNi4−x-3OAc-ClO3
was synthesized from 0.7 mmol Co(OAc)2 + 0.1 mmol
Ni(OAc)2 and 3 mmol NaClO3. Additionally, this product
further expanded the spectrum of type 1 {CoxNi4−xO4}
cubanes (Figure S66b and S67).
Mixed Co/Ni Pathways. In summary, mixed type 2 cubanes
can generally be accessed from precursors containing ca. ≥25
mol % Ni2+ cations due to their strong preference for the
CONCLUSIONS
■
We introduce an unprecedented controlled access to a large
family of {M4O4} (M = Co, Ni, and Co/Ni) bioinspired
cubane photocatalysts. A toolbox of straightforward inorganic
I
DOI: 10.1021/jacs.9b01356
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
counteranions directs the assembly of the di(2-pyridyl) ketone
(dpk) ligand and M(OAc)2 precursors in aqueous solution into
clearly accessible cubane types:
Karrar Al-Ameed: 0000-0002-8333-3868
Mauro Schilling: 0000-0002-9394-4726
Thomas Fox: 0000-0002-7572-8210
Sandra Luber: 0000-0002-6203-9379
Greta R. Patzke: 0000-0003-4616-7183
Type 2 {Co4O4} cubanes with the characteristic {H2O-
Co2(OR)2-OH2} edge-site motif are selectively obtained with
perchlorate anions. In contrast, a wide range of counteranions
−
−
−
−
−
(≤0.5 equiv of ClO4 , and ClO3 , NO3 , BF4 , Cl−, PF6 )
provide access to type 1 Co-cubanes with adjustable ligand
patterns. Type 2 edge-site {Ni4O4} cubanes are formed with
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
−
■
>0.3 equiv ClO4 , while all other tested counteranions provide
This work has been supported by the University of Zurich, the
University Research Priority Program (URPP) for Solar Light
to Chemical Energy Conversion (LightChEC), and by the
Swiss National Science Foundation (Sinergia Grant No.
CRSII2_160801/1). S. Luber thanks the Swiss National
Science Foundation (Grant No. PP00P2_170667) for financial
support. K. Al-Ameed thanks the Swiss Government
Excellence Scholarships for Foreign Scholars and Artists for
financial support. We thank Prof. Dr. Anthony Linden and Dr.
Olivier Blacque (Department of Chemistry, University of
Zurich) for helpful crystallographic discussions. We are grateful
to PD Dr. Laurent Bigler (Department of Chemistry,
University of Zurich) for support and discussions concerning
HR-ESI-MS measurements.
2-μ-OAc-Ni4. The 2-CoxNi4−x and 2-μ-OAc-CoxNi4−x cu-
banes can be tailored via counteranion controlled assembly of
mixed precursors. Furthermore, the homogeneous visible-light-
driven water oxidation activity of type 1 and type 2 {Co4O4}
cubanes with increasingly accessible edge-site regions was
compared.
Systematic screenings provided cumulative experimental
evidence for type 2 and 1 {Co4O4} cubane assembly via two
main building blocks, namely, [Co2(dpy-C{OH}O)2(H2O)4]2+
and [Co2(dpy-C{OH}O)2(OAc)2(H2O)2]. DFT calculations
confirmed that [Co2(dpy-C{OH}O)2(OAc)2(H2O)2] is in-
deed thermodynamically most stable of all investigated dimer
types in solution. Computed reaction energies agree with the
experimentally supported formation pathway of type 1 cubane
1-4OAc through a head-to-head combination of these dimers.
Furthermore, DFT results indicate that type 2 cubanes arise
from the reaction of the [Co2(dpy-C{OH}O)2(OAc)2(H2O)2]
dimer with the most stable [Co(dpy-C{OH}O)(OAc)-
(H2O)3] monomer that is likely to result from higher applied
perchlorate strengths. The selective access to type 2 {Ni4O4}
and type 2 mixed {(Co/Ni)4O4} cubanes proceeds via
combinations of according Ni-dimers with predominant
[Ni2+] species through analogous assembly processes.
REFERENCES
■
(1) Ryan, K.; Lengyel, J.; Shatruk, M. Crystal Structure Prediction
via Deep Learning. J. Am. Chem. Soc. 2018, 140, 10158−10168.
(2) Coley, C. W.; Green, W. H.; Jensen, K. F. Machine Learning in
Computer-Aided Synthesis Planning. Acc. Chem. Res. 2018, 51, 1281−
1289.
(3) Kim, E.; Huang, K.; Jegelka, S.; Olivetti, E. Virtual screening of
inorganic materials synthesis parameters with deep learning. NPJ
Comput. Mater. 2017, 3, 1120.
We reveal for the first time the high potential of convenient
inorganic counteranions as structure-directing agents for oxo
cluster assembly in solution. The formation pathways leading
to an arsenal of transition metal cubanes were tracked with an
effective combination of mechanistic screening going hand in
hand with DFT calculations. Our extensive data set offers an
excellent platform the for forthcoming machine learning
explorations of the full parameter space in coordination
chemistry. We here first pave the way to facile, predictive,
and mechanistically driven design concepts for the strategic
construction of functional transition metal complexes.
(4) Bijelic, A.; Aureliano, M.; Rompel, A. Polyoxometalates as
Potential Next-Generation Metallodrugs in the Combat Against
Cancer. Angew. Chem., Int. Ed. 2019, 57, 2−22.
(5) Genovese, M.; Lian, K. Polyoxometalate modified inorganic−
organic nanocomposite materials for energy storage applications: A
review. Curr. Opin. Solid State Mater. Sci. 2015, 19, 126−137.
(6) Xu, G.; Wang, Z.; Ling, R.; Zhou, J.; Chen, X.-D.; Holm, R. H.
Ligand metathesis as rational strategy for the synthesis of cubane-type
heteroleptic iron-sulfur clusters relevant to the FeMo cofactor. Proc.
Natl. Acad. Sci. U. S. A. 2018, 115, 5089−5092.
(7) Luo, Z.; Zhou, M.; Wang, X. Cobalt-based cubane molecular co-
catalysts for photocatalytic water oxidation by polymeric carbon
nitrides. Appl. Catal., B 2018, 238, 664−671.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.9b01356.
(8) Schwarz, B.; Forster, J.; Goetz, M. K.; Yucel, D.; Berger, C.;
̈
■
Jacob, T.; Streb, C. Visible-Light-Driven Water Oxidation by a
Molecular Manganese Vanadium Oxide Cluster. Angew. Chem., Int.
Ed. 2016, 55, 6329−6333.
S
(9) Maayan, G.; Gluz, N.; Christou, G. A Bioinspired Soluble
Manganese Cluster as a Water Oxidation Electrocatalyst with Low
Overpotential. Nat. Catal. 2018, 1, 48−54.
Detailed synthetic and catalytic protocols, stability tests,
analytical investigations, and computational methods
and results (PDF)
Crystal data (also deposited as CCDC 1891281-
1891291, 1891332-1891335, 1891347-1891350,
1891352-1891357, 1891359-1891361, and 1891379-
1891384) (CIF)
(10) Han, X.-B.; Li, Y.-G.; Zhang, Z.-M.; Tan, H.-Q.; Lu, Y.; Wang,
E.-B. Polyoxometalate-based nickel clusters as visible light-driven
water oxidation catalysts. J. Am. Chem. Soc. 2015, 137, 5486−5493.
(11) Han, X.-B.; Zhang, Z.-M.; Zhang, T.; Li, Y.-G.; Lin, W.; You,
W.; Su, Z.-M.; Wang, E.-B. Polyoxometalate-based cobalt-phosphate
molecular catalysts for visible light-driven water oxidation. J. Am.
Chem. Soc. 2014, 136, 5359−5366.
AUTHOR INFORMATION
Corresponding Author
*greta.patzke@chem.uzh.ch
(12) Lant, H. M. C.; Michaelos, T. K.; Sharninghausen, L. S.;
Mercado, B. Q.; Crabtree, R. H.; Brudvig, G. W. N, N, O Pincer
Ligand with a Deprotonatable Site that Promotes Redox-Leveling,
High Mn Oxidation States, and a Mn2O2 Dimer Competent for
Catalytic Oxygen Evolution. Eur. J. Inorg. Chem. 2019, 2019, 2115−
2123.
■
ORCID
Fangyuan Song: 0000-0002-9563-1926
J
DOI: 10.1021/jacs.9b01356
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
́
(13) Misra, A.; Franco Castillo, I.; Mu
̈
ller, D. P.; Gonzalez, C.;
(29) Hu, J.; Zhang, J.; Zhao, J. A.; Hu, L.; Chen, S. Anion-driven
Ag(I) coordination architectures: Structural diversity and photo-
luminescence behaviors. J. Coord. Chem. 2016, 69, 574−584.
(30) Liu, Y.; Li, J.; Hou, H.; Fan, Y. Subtle role of counteranions in
molecular construction: Structures and properties of novel Cu(II)
coordination complexes with bis-(1-benzoimidazolymethylene)-(2,5-
thiadiazoly)-disulfide. J. Organomet. Chem. 2009, 694, 2875−2882.
(31) Yang, D.; Zhao, J.; Yang, X.-J.; Wu, B. Anion-coordination-
directed self-assemblies. Org. Chem. Front. 2018, 5, 662−690.
(32) Zhao, J.; Yang, D.; Yang, X.-J.; Wu, B. Anion coordination
chemistry: From recognition to supramolecular assembly. Coord.
Chem. Rev. 2019, 378, 415−444.
Eyssautier-Chuine, S.; Ziegler, A.; La Fuente, J. M. de; Mitchell, S. G.;
Streb, C. Polyoxometalate-Ionic Liquids (POM-ILs) as Anticorrosion
and Antibacterial Coatings for Natural Stones. Angew. Chem., Int. Ed.
2018, 57, 14926−14931.
(14) Zhang, Y.; Zhang, J.-J.; Jia, A.-Q.; Xin, Z.-F.; Zhang, Q.-F. A
Single-Molecule Magnet Tetranuclear [Mn3IIIMnIVO3Cl] Complex
with Bis(diisopropylphosphinyl)imide Ligands. J. Cluster Sci. 2018,
29, 1345−1352.
(15) Lang, Z.-L.; Guan, W.; Wu, Z.-J.; Yan, L.-K.; Su, Z.-M. Building
blocks and formation thermodynamics of α-Keggin-type
[PW12O40]3- anion. Comput. Theor. Chem. 2012, 999, 66−73.
(16) Demissie, T. B.; Ruud, K.; Hansen, J. H. DFT as a Powerful
Predictive Tool in Photoredox Catalysis: Redox Potentials and
Mechanistic Analysis. Organometallics 2015, 34, 4218−4228.
(17) Tong, Y.-P.; Luo, G.-T.; Zhen, J.; Shen, Y.; Lin, Y.-W.
Synthesis, structures, and theoretical investigation of three poly-
oxomolybdate-based compounds: Self-assembly, fragment analysis,
orbital interaction, and formation mechanism. CrystEngComm 2015,
17, 2629−2635.
(33) Muller, M.; Albrecht, M.; Sackmann, J.; Hoffmann, A.; Dierkes,
̈
F.; Valkonen, A.; Rissanen, K. CH-anion versus anion-π interactions
in the crystal and in solution of pentafluorobenzyl phosphonium salts.
Dalton Trans. 2010, 39, 11329−11334.
(34) Das, A.; Choudhury, S. R.; Estarellas, C.; Dey, B.; Frontera, A.;
Hemming, J.; Helliwell, M.; Gamez, P.; Mukhopadhyay, S. Supra-
molecular assemblies involving anion−π and lone pair−π interactions:
Experimental observation and theoretical analysis. CrystEngComm
2011, 13, 4519.
(18) Thomas, P.; Chandel, S.; Mallick, A.; Sreejith, S. S.; Ghosh, N.;
Roy, S. Studying the Crystallization of Polyoxometalates from
Colloidal Softoxometalates. Cryst. Growth Des. 2018, 18, 4068−4075.
(19) Wong, H.-Y.; Chan, W. T. K.; Law, G.-L. Assembly of
Lanthanide(III) Cubanes and Dimers with Single-Molecule Magnet-
ism and Photoluminescence. Inorg. Chem. 2018, 57, 6893−6902.
(20) Michaelos, T. K.; Shopov, D. Y.; Sinha, S. B.; Sharninghausen,
L. S.; Fisher, K. J.; Lant, H. M. C.; Crabtree, R. H.; Brudvig, G. W. A
Pyridine Alkoxide Chelate Ligand That Promotes Both Unusually
High Oxidation States and Water-Oxidation Catalysis. Acc. Chem. Res.
2017, 50, 952−959.
(21) Shopov, D. Y.; Sharninghausen, L. S.; Sinha, S. B.; Borowski, J.
E.; Mercado, B. Q.; Brudvig, G. W.; Crabtree, R. H. Synthesis of
Pyridine-Alkoxide Ligands for Formation of Polynuclear Complexes.
New J. Chem. 2017, 41, 6709−6719.
(22) Libby, E.; Folting, K.; Huffman, C. J.; Huffman, J. C.; Christou,
G. Controlled aggregation and deaggregation of a polynuclear oxido-
bridged manganese complex: utility to synthesis of the Jahn-Teller
effect in high-spin manganese(III). Inorg. Chem. 1993, 32, 2549−
2556.
(35) Gale, P. A.; Quesada, R. Anion coordination and anion-
templated assembly: Highlights from 2002 to 2004. Coord. Chem. Rev.
2006, 250, 3219−3244.
(36) Cui, K.; Mali, K. S.; Wu, D.; Feng, X.; Mullen, K.; Walter, M.;
̈
Feyter, S. de; Mertens, S. F. L. Reversible Anion-Driven Switching of
an Organic 2D Crystal at a Solid-Liquid Interface. Small 2017, 13,
1702379.
(37) Srivastava, A. K.; Praveenkumar, B.; Mahawar, I. K.; Divya, P.;
Shalini, S.; Boomishankar, R. Anion Driven [CuIIL2]n Frameworks:
Crystal Structures, Guest-Encapsulation, Dielectric, and Possible
Ferroelectric Properties. Chem. Mater. 2014, 26, 3811−3817.
(38) Oliva, L.; Oliva, P.; Galdi, N.; Pellecchia, C.; Sian, L.;
Macchioni, A.; Zuccaccia, C. Solution Structure and Reactivity with
Metallocenes of AlMe2 F: Mimicking Cation-Anion Interactions in
Metallocenium-Methylalumoxane Inner-Sphere Ion Pairs. Angew.
Chem., Int. Ed. 2017, 56, 14227−14231.
(39) Ciancaleoni, G.; Belpassi, L.; Tarantelli, F.; Zuccaccia, D.;
Macchioni, A. A combined NMR/DFT study on the ion pair structure
of (PR1(2)R2)Au(η2−3-hexyne)BF4 complexes. Dalton Trans. 2013,
42, 4122−4131.
(40) Zheng, X.-Y.; Xie, J.; Kong, X.-J.; Long, L.-S.; Zheng, L.-S.
Recent advances in the assembly of high-nuclearity lanthanide
clusters. Coord. Chem. Rev. 2019, 378, 222−236.
(41) Healy, C.; Schmitt, W. Multicomponent halide templating: The
effect of structure-directing agents on the assembly of molecular and
extended coordination compounds. Coord. Chem. Rev. 2018, 371, 67−
85.
(42) Trinapakul, M.; Kritayakornupong, C.; Tongraar, A.;
Vchirawongkwin, V. Active site of the solvated thiosulfate ion
characterized by hydration structures and dynamics. Dalton Trans.
2013, 42, 10807−10817.
(43) Eklund, L.; Hofer, T. S.; Pribil, A. B.; Rode, B. M.; Persson, I.
On the structure and dynamics of the hydrated sulfite ion in aqueous
solution–an ab initio QMCF MD simulation and large angle X-ray
scattering study. Dalton Trans. 2012, 41, 5209−5216.
(44) Eklund, L.; Hofer, T. S.; Weiss, A. K. H.; Tirler, A. O.; Persson,
I. Structure and water exchange of the hydrated thiosulfate ion in
aqueous solution using QMCF MD simulation and large angle X-ray
scattering. Dalton Trans. 2014, 43, 12711−12720.
(23) Nguyen, A. I.; Darago, L. E.; Balcells, D.; Tilley, T. D. Influence
of a “Dangling” Co(II) Ion Bound to a MnCo3O4 Oxo Cubane. J.
Am. Chem. Soc. 2018, 140, 9030−9033.
(24) Akhtar, M. N.; Lan, Y.; AlDamen, M. A.; Zheng, Y.-Z.; Anson,
C. E.; Powell, A. K. Effect of ligand substitution on the SMM
properties of three isostructural families of double-cubane Mn4Ln2
coordination clusters. Dalton Trans. 2018, 47, 3485−3495.
(25) Modak, R.; Sikdar, Y.; Thuijs, A. E.; Christou, G.; Goswami, S.
CoII4, CoII7, and a Series of CoII2LnIII (LnIII = NdIII, SmIII,
GdIII, TbIII, DyIII) Coordination Clusters: Search for Single
Molecule Magnets. Inorg. Chem. 2016, 55, 10192−10202.
̀
(26) Christie, L. G.; Asche, S.; Mathieson, J. S.; Vila-Nadal, L.;
Cronin, L. Investigating the Formation of Giant {Pd72}Prop and
{Pd84}Gly Macrocycles Using NMR, HPLC, and Mass Spectrometry.
J. Am. Chem. Soc. 2018, 140, 9379−9382.
(27) Tandon, S. S.; Bunge, S. D.; Rakosi, R.; Xu, Z.; Thompson, L.
K. Self-assembly of mixed-valence Co(II/III) and Ni(II) clusters:
azide-bridged 1D single chain coordination polymers comprised of
tetranuclear units, tetranuclear Co(II/III) complexes, ferromagneti-
cally coupled azide-bridged tetranuclear, and hexanuclear Ni(II)
complexes: synthesis, structural, and magnetic properties. Dalton
Trans. 2009, 6536−6551.
(45) Eklund, L.; Hofer, T. S.; Persson, I. Structure and water
exchange dynamics of hydrated oxo halo ions in aqueous solution
using QMCF MD simulation, large angle X-ray scattering and EXAFS.
Dalton Trans. 2015, 44, 1816−1828.
́
́
́
̌
́
(28) Hollo, B.; Rodio, M. V.; Bera, O.; Jovicio, M.; Leovac, V. M.;
(46) Smiechowski, M. Anion−water interactions of weakly hydrated
́
́
́
́
́
Tomio, Z. D.; Meszaros Szecsenyi, K. Cation- and/or anion-directed
reaction routes. Could the desolvation pattern of isostructural
coordination compounds be related to their molecular structure?
Struct. Chem. 2013, 24, 2193−2201.
anions: Molecular dynamics simulations of aqueous NaBF4 and
NaPF6. Mol. Phys. 2016, 114, 1831−1846.
(47) Elmehdawi, R. M.; El-Kaheli, M. N.; Abuhmaiera, R. G.; Treish,
F. A.; Ben Younes, M. E. M.; Bazzicalupi, C.; Guerri, A.; Caneschi, A.;
K
DOI: 10.1021/jacs.9b01356
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Amjad, A. Synthesis, Crystal Structure, and Magnetic Properties of a
New Mixed Metal (Co(II), Ni(II)) Cubane. Materials 2017, 10, 178.
(48) Gahan, L. R.; Henriksen, S. L. Preparation and characterization
of di-(μ2-CH3COO-κ2O: O′)-tetrakis[μ3-methoxo-2,4-
pentanedionatometal(II, III)] complexes. Polyhedron 2015, 95, 30−
39.
(49) Malpaharia, P.; Pramanik, K.; Costes, J.-P.; Tuchagues, J.-P.;
Moulton, B.; Zaworotko, M. J.; Das, B.; Chandra, S. K. Tetranuclear
[Mn 2 Co 2 ], [Mn 2 Fe 2 ], and [Mn 2 Mn 2 ] Complexes with
Defective Double-Cubane Cores and Phenoxo and Oxo Bridges:
Syntheses, Crystal Structures, and Electronic Properties. Eur. J. Inorg.
Chem. 2014, 2014, 3527−3535.
(50) Zhang, S.-H.; Zhao, R.-X.; Li, G.; Zhang, H.-Y.; Zhang, C.-L.;
Muller, G. Structural variation from heterometallic heptanuclear or
heptanuclear to cubane clusters based on 2-hydroxy-3-ethoxy-
benzaldehyde: Effects of pH and temperature. RSC Adv. 2014, 4,
54837−54846.
(51) Kessler, V. G.; Gohil, S.; Kritikos, M.; Korsak, O. N.; Knyazeva,
E. E.; Moskovskaya, I. F.; Romanovsky, B. V. An approach to
heterometallic alkoxide-β-diketonate complexes with a M4O4 cubane-
like core and new prospects of their application in preparation of solid
catalysts. X-ray single crystal study of (Co, Ni)4(acac)4(μ3-OMe)-
4(MeOH)4, Co2Ni2(acac)4(μ3-OMe)4(OAc)2 and Mg4(acac)4-
(μ3-OMe)4(MeOH)4. Polyhedron 2001, 20, 915−922.
(63) Schilling, M.; Hodel, F. H.; Luber, S. Discovery of Open
Cubane Core Structures for Biomimetic LnCo3(OR)4 Water
Oxidation Catalysts. ChemSusChem 2017, 10, 4561−4569.
(64) Kanady, J. S.; Tsui, E. Y.; Day, M. W.; Agapie, T. A Synthetic
Model of the Mn3Ca Subsite of the Oxygen-Evolving Complex in
Photosystem II. Science 2011, 333, 733−736.
(65) Zhang, C.; Chen, C.; Dong, H.; Shen, J.-R.; Dau, H.; Zhao, J. A
synthetic Mn4Ca-cluster mimicking the oxygen-evolving center of
photosynthesis. Science 2015, 348, 690−693.
̌
(66) Efthymiou, C. G.; Raptopoulou, C. P.; Terzis, A.; Boca, R.;
Korabic, M.; Mrozinski, J.; Perlepes, S. P.; Bakalbassis, E. G. A
Systematic Exploration of Nickel(II)/Acetate/Di-2-pyridyl Ketone
Chemistry: Neutral and Cationic Tetranuclear Clusters, and a Novel
Mononuclear Complex. Eur. J. Inorg. Chem. 2006, 2006, 2236−2252.
(67) Li, Y.-M.; Zhang, J.-J.; Fu, R.-B.; Xiang, S.-C.; Sheng, T.-L.;
Yuan, D.-Q.; Huang, X.-H.; Wu, X.-T. Three new cubane-like
transition metal complexes of di-2-pyridyl ketone in gem-diol form:
Syntheses, crystal structures and properties. Polyhedron 2006, 25,
1618−1624.
(68) Padhi, S. K.; Sahu, R. Co(II/III) coordinated pyridine
alcoholate ligand generated through metal assisted nucleophilic
addition to a CO function: Temperature dependent synthesis of a
mononuclear complex and a neutral cubane cluster. Polyhedron 2008,
27, 2662−2666.
(69) Papatriantafyllopoulou, C.; Efthymiou, C. G.; Raptopoulou, C.
P.; Vicente, R.; Manessi-Zoupa, E.; Psycharis, V.; Escuer, A.; Perlepes,
S. P. Initial use of the di-2-pyridyl ketone/sulfate “blend” in 3d-metal
cluster chemistry: Preparation, X-ray structures and physical studies of
zinc(II) and nickel(II) cubanes. J. Mol. Struct. 2007, 829, 176−188.
(70) Proll, P. J.; Sutcliffe, L. H.; Walkley, J. Species of Cobalt(II) in
Acetic Acid. Part I. Cobaltous Acetate in the Presence of Water and of
Sodium Acetate. J. Phys. Chem. 1961, 65, 455−460.
(71) Bardan, S.; Aditya, S. J. Indian. Chem. Soc. 1955, 32, 109.
(72) Katsenis, A. D.; Inglis, R.; Slawin, A. M. Z.; Kessler, V. G.;
Brechin, E. K. Transforming the cube: a tetranuclear cobalt(II)
cubane cluster and its transformation to a dimer of dimers.
CrystEngComm 2009, 11, 2117−2120.
(73) Katsoulakou, E.; Lalioti, N.; Raptopoulou, C. P.; Terzis, A.;
Manessi-Zoupa, E.; Perlepes, S. P. Insights into the role of zinc(II)
sites in hydrolytic enzymes: Study of the ZnII/X/(py)2CO (X = Cl-,
N3-, SO42-;) reaction systems. Inorg. Chem. Commun. 2002, 5, 719−
723.
(52) Lin, P.-H.; Tsui, E. Y.; Habib, F.; Murugesu, M.; Agapie, T.
Effect of the Mn Oxidation State on Single-Molecule-Magnet
Properties: Mn(III) vs Mn(IV) in Biologically Inspired DyMn3O4
Cubanes. Inorg. Chem. 2016, 55, 6095−6099.
(53) Kanady, J. S.; Lin, P.-H.; Carsch, K. M.; Nielsen, R. J.; Takase,
M. K.; Goddard, W. A.; Agapie, T. Toward models for the full oxygen-
evolving complex of photosystem II by ligand coordination to lower
the symmetry of the Mn3CaO4 cubane: demonstration that
electronic effects facilitate binding of a fifth metal. J. Am. Chem. Soc.
2014, 136, 14373−14376.
́
(54) Evangelisti, F.; Guttinger, R.; More, R.; Luber, S.; Patzke, G. R.
̈
Closer to photosystem II: a Co4O4 cubane catalyst with flexible
ligand architecture. J. Am. Chem. Soc. 2013, 135, 18734−18737.
́
(55) Evangelisti, F.; More, R.; Hodel, F.; Luber, S.; Patzke, G. R. 3d-
4f {Co(II)3Ln(OR)4} Cubanes as Bio-Inspired Water Oxidation
Catalysts. J. Am. Chem. Soc. 2015, 137, 11076−11084.
́
(56) Song, F.; More, R.; Schilling, M.; Smolentsev, G.; Azzaroli, N.;
Fox, T.; Luber, S.; Patzke, G. R. {Co4O4} and {CoxNi4-xO4}
Cubane Water Oxidation Catalysts as Surface Cut-Outs of Cobalt
Oxides. J. Am. Chem. Soc. 2017, 139, 14198−14208.
(57) Olshansky, L.; Huerta-Lavorie, R.; Nguyen, A. I.; Vallapurackal,
J.; Furst, A.; Tilley, T. D.; Borovik, A. S. Artificial Metalloproteins
Containing Co4O4 Cubane Active Sites. J. Am. Chem. Soc. 2018, 140,
2739−2742.
(74) Bickley, R. I.; Edwards, H. G. M.; Rose, S. J.; Gustar, R. A
raman spectroscopic study of nickel(II) acetate, Ni(CH3COO)2 and
its aqueous and methanolic solutions. J. Mol. Struct. 1990, 238, 15−
26.
(75) La Ganga, G.; Nardo, V. M.; Cordaro, M.; Natali, M.; Vitale, S.;
Licciardello, A.; Nastasi, F.; Campagna, S. A functionalized, ethynyl-
decorated, tetracobalt(III) cubane molecular catalyst for photo-
induced water oxidation. Dalton Trans. 2014, 43, 14926−14930.
(76) Berardi, S.; La Ganga, G.; Natali, M.; Bazzan, I.; Puntoriero, F.;
Sartorel, A.; Scandola, F.; Campagna, S.; Bonchio, M. Photocatalytic
water oxidation: tuning light-induced electron transfer by molecular
Co4O4 cores. J. Am. Chem. Soc. 2012, 134, 11104−11107.
(77) McCool, N. S.; Robinson, D. M.; Sheats, J. E.; Dismukes, G. C.
A Co4O4 “cubane” water oxidation catalyst inspired by photosyn-
thesis. J. Am. Chem. Soc. 2011, 133, 11446−11449.
(58) Nguyen, A. I.; Wang, J.; Levine, D. S.; Ziegler, M. S.; Tilley, T.
D. Synthetic control and empirical prediction of redox potentials for
Co4O4 cubanes over a 1.4 V range: implications for catalyst design
and evaluation of high-valent intermediates in water oxidation. Chem.
Sci. 2017, 8, 4274−4284.
(59) Nguyen, A. I.; Ziegler, M. S.; Ona-Burgos, P.; Sturzbecher-
̃
Hohne, M.; Kim, W.; Bellone, D. E.; Tilley, T. D. Mechanistic
Investigations of Water Oxidation by a Molecular Cobalt Oxide
Analogue: Evidence for a Highly Oxidized Intermediate and Exclusive
Terminal Oxo Participation. J. Am. Chem. Soc. 2015, 137, 12865−
12872.
(78) Kirchhoff, B.; Rau, S.; Streb, C. Detecting and Preventing the
Formation of Photosensitizer-Catalyst Colloids in Homogeneous
Light-Driven Water Oxidation. Eur. J. Inorg. Chem. 2016, 2016,
1425−1429.
(60) Liao, R.-Z.; Siegbahn, P. E. M. Possible water association and
oxidation mechanisms for a recently synthesized Mn4Ca-complex. J.
Catal. 2017, 354, 169−181.
(61) Hodel, F. H.; Luber, S. Redox-Inert Cations Enhancing Water
Oxidation Activity: The Crucial Role of Flexibility. ACS Catal. 2016,
6, 6750−6761.
(62) Hodel, F. H.; Luber, S. What Influences the Water Oxidation
Activity of a Bioinspired Molecular CoII4O4 Cubane? An In-Depth
Exploration of Catalytic Pathways. ACS Catal. 2016, 6, 1505−1517.
L
DOI: 10.1021/jacs.9b01356
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX