Water Oxidation Catalysis by Synthetic Manganese Oxides with Different Structural Motifs: A Comparative Study

Article abstract of DOI:10.1002/chem.201501367

Manganese oxides are considered to be very promising materials for water oxidation catalysis (WOC), but the structural parameters influencing their catalytic activity have so far not been clearly identified. For this study, a dozen manganese oxides (MnO_x) with various solid-state structures were synthesised and carefully characterised by various physical and chemical methods. WOC by the different MnO_x was then investigated with Ce⁴⁺ as chemical oxidant. Oxides with layered structures (birnessites) and those containing large tunnels (todorokites) clearly gave the best results with reaction rates exceeding 1250 mmol_O2 mol^-1_Mn h^-1 or about 50μmol_O2m^-2h^-1. In comparison, catalytic rates per mole of Mn of oxides characterised by well-defined 3D networks were rather low (e.g., ca.90mmol_O2mmol^-1h^-1 for bixbyite, Mn₂O₃), but impressive if normalised per unit surface area (>100μmol^-2h^-1 for marokite, CaMn₂O₄). Thus, two groups of MnO_x emerge from this screening as hot candidates for manganese-based WOC materials: 1)amorphous oxides with tunnelled structures and the well-established layered oxides; 2)crystalline Mn^III oxides. However, synthetic methods to increase surface areas must be developed for the latter to obtain good catalysis rates per mole of Mn or per unit catalyst mass. WOC this way: A detailed, comparative study on twelve synthetic manganese oxides revealed that water oxidation catalysis is greatly influenced by the oxide structure (see figure). Layered oxides and those featuring large tunnels performed best, whereas crystalline 3D-network materials showed little activity.

Full text of DOI:10.1002/chem.201501367

DOI: 10.1002/chem.201501367
Full Paper
&
Manganese Oxides |Hot Paper|
Water Oxidation Catalysis by Synthetic Manganese Oxides with
Different Structural Motifs: A Comparative Study
Carolin E. Frey and Philipp Kurz*[a]
Abstract: Manganese oxides are considered to be very
promising materials for water oxidation catalysis (WOC), but
the structural parameters influencing their catalytic activity
have so far not been clearly identified. For this study,
a dozen manganese oxides (MnOx) with various solid-state
structures were synthesised and carefully characterised by
various physical and chemical methods. WOC by the differ-
ent MnOx was then investigated with Ce4+ as chemical oxi-
dant. Oxides with layered structures (birnessites) and those
containing large tunnels (todorokites) clearly gave the best
mole of Mn of oxides characterised by well-defined 3D net-
works were rather low (e.g., ca. 90 mmolO molÀM1n hÀ1 for
2
bixbyite, Mn2O3), but impressive if normalised per unit sur-
face area (>100 mmolO
m
À2 hÀ1 for marokite, CaMn2O4).
2
Thus, two groups of MnOx emerge from this screening as
hot candidates for manganese-based WOC materials:
1) amorphous oxides with tunnelled structures and the well-
established layered oxides; 2) crystalline MnIII oxides. Howev-
er, synthetic methods to increase surface areas must be de-
veloped for the latter to obtain good catalysis rates per
mole of Mn or per unit catalyst mass.
results with reaction rates exceeding 1250 mmolO molÀM1n hÀ1
2
or about 50 mmolO2 m
À2 hÀ1. In comparison, catalytic rates per
Introduction
In artificial, bio-inspired systems for water splitting, manga-
nese oxides (MnOx) have been used as materials for water oxi-
dation catalysis (WOC) since the late 1960s.[6–8] This approach
has been “rediscovered” in recent years, and it has been found
that good WOC rates can especially be obtained if MnOx from
the so-called birnessite family of oxides are used as heteroge-
neous, Mn-based catalyst materials.[8–13] Birnessites are manga-
nese oxides containing layers of [MnO6] octahedra. The inter-
layer distance is approximately 7 Š and the average Mn oxida-
tion state is in the range of +3.5–3.9.[14–16] Varying amounts of
water and secondary cations (especially of group 1 and 2
metals) can be incorporated into the interlayer space, and in
consequence a large variety of birnessites with different chemi-
cal compositions are synthetically accessible.[8,15] We and
a number of other groups have used this compositional flexi-
bility of birnessite-type MnOx materials in recent years to opti-
mise their WOC activities, and significant progress has been
made, especially by variations of oxide compositions, sintering
temperatures and particle sizes of synthetic birnessites. Howev-
er, a large number of other manganese oxides with different
solid-state structures is known (more than 30 types have been
described;[14] see Figure 1 for examples), but differences in
WOC activities between these oxide types have rarely been
studied. Hence, in this work, we prepared and investigated
a dozen MnOx materials with different compositions and struc-
tures in order to establish trends in catalytic activities. By
doing so, we hoped to find new leads for promising MnO cata-
lysts, as well as to obtain a better understanding of why bir-
nessites have so far dominated the field of Mn-based WOC.
A range of different types of manganese oxides have already
been tested for WOC, albeit rarely under identical experimental
The future of the world’s energy supply is renewable energy.
In 2012, an estimated 19% of the world’s energy consumption
was renewable energy, and the prediction for 2013 was even
higher.[1] A disadvantage of most renewable energy sources is
that production cannot easily be regulated to meet demand.
Therefore, technologies to store large amounts of renewable
energy are needed. The splitting of water into hydrogen and
oxygen is one of the prominent concepts for this task, since
the produced H2 could be used as a storable and environmen-
tally friendly fuel. To make this reaction economically feasible,
a central problem to be solved is the development of an af-
fordable, abundant and non-toxic catalyst for the oxidation of
water to dioxygen, a reaction that is central to all such renewa-
ble-fuel schemes.
In biology, water oxidation takes place at the oxygen-evolv-
ing complex (OEC) of the photosystem II (PSII) enzyme com-
plex.[2] The OEC consist of a CaMn4O5 cluster embedded in an
amino acid environment.[2,3] The metal ions of the OEC are in-
terconnected by m-oxido bridges, and the Mn oxidation states
switch between Mn3+ and Mn4+ in the catalytic cycle, which
involves four single-electron transfer steps.[4,5]
[a] C. E. Frey, Prof. Dr. P. Kurz
Albert-Ludwigs-Universität Freiburg
Institut für Anorganische und Analytische Chemie
Albertstrasse 21, 79106 Freiburg (Germany)
E-mail: philipp.kurz@ac.uni-freiburg.de
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201501367.
Chem. Eur. J. 2015, 21, 14958 – 14968
14958
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
building blocks (Figure 1). The first group of manganese oxides
can be summarised as crystalline, 3D networks of Mn2+/3+/4+
and O2À ions. Bixbyite (a-Mn2O3), hausmannite (Mn3O4) and
LiMn2O4 (both spinel-like oxides), marokite (CaMn2O4) as well
as manganosite (MnO, NaCl structure) are typical representa-
tives of this group.
In the second structural type, [MnO6] octahedra are arranged
in the form of tunnels. These tunnelled oxides can be subdivid-
ed by their tunnel architectures (e.g., [2”1], [3”3]; see
Figure 1), which also influence the amount of secondary cat-
ions and/or water that can be incorporated into the tunnelled
structures. The average oxidation state of manganese here is
usually between +3.0 and +4.0, and pyrolusite, ramsdellite,
hollandite, cryptomelane and todorokite are common exam-
ples of this class.
A third class of manganese oxides are layered minerals.
Prominent examples are birnessites, which form a large sub-
group of compounds in the class of layered Mn oxides, and
vernadite (d-MnO2). Layered MnOx materials tend to be highly
amorphous, and so details about their structures are often not
known at the atomic level. However, common features of lay-
ered manganese oxides are interlayer distances of 7–10 Š, sig-
nificant amounts of intercalated secondary cations and water
and average manganese oxidation states of +3.5 to +4.0.
Given this large variety of solid-state structures and the
known WOC activity of some manganese oxides, different fac-
tors have been suggested to be important for the catalytic
process: 1) similar to the OEC, flexible structures should be
beneficial for substrate binding and oxidation-state
changes,[5,9,10,12,18] 2) high surface areas result in good accessi-
bility of the catalytically active centres,[7,12,13,19,20,25] 3) average
Mn oxidation states greater than +3.5 often provide enough
thermodynamic driving force for water oxidation,[7,18,22,26]
4) secondary cations such as (Lewis acidic) Ca2+ assist water
oxidation by activation of the H2O substrate[10,12,13] and 5) good
conductivity of the oxide enhances performance, because
WOC requires that four oxidation equivalents accumulate at
the (so far unidentified) catalytic sites in the MnOx.[9,18] Some of
these factors are obviously contradictory to each other, and so
there is currently no real consensus on the most important
structural parameters influencing WOC activities of manganese
oxides.
Figure 1. Structural motifs of some common manganese oxides.[14] Networks
(black labels): a) bixbyite (Mn2O3); b) spinel-type (hausmannite (Mn3O4,
LiMn2O4); c) marokite (CaMn2O4); d) manganosite (MnO). Tunnelled struc-
tures (grey labels): e) pyrolusite, (MnO2); f) ramsdellite; g) hollandite-group;
h) todorokite. Layered oxides (light grey labels): i) birnessite group. Colour
code: black=oxide (O2À) or O of OHÀ/H2O; grey octahedra/tetrahedral =
manganese; white=calcium or magnesium; light grey=potassium or
sodium.
conditions.[6,8,12,13,17–23] Instead, catalyst concentrations, oxidis-
ing agents, pH regime and so forth differed greatly from study
to study, and the reported WOC rates therefore can hardly be
compared. To our knowledge, in only one report to date have
WOC rates of more than three different types of manganese
oxide been screened in the same catalysis set-up, namely, the
report by Robinson et al. in 2013 on a set of crystalline MnOx
compounds and their WOC performances.[20] Robinson et al.
first determined the identity of the synthetic oxides by XRD,
analysed the surface areas by gas-sorption experiments and
then screened catalytic activities with photogenerated
[Ru(bpy)3]3+ (bpy=2,2’-bipyridine) as oxidising agent. Reaction
rates were normalised to catalyst surface area, and bixbyite
(Mn2O3) was found to be the most active catalyst. In compari-
son, slow WOC was observed for hausmannite (Mn3O4) and
synthetic l-MnO2 (a spinel-type oxide). Five other oxides were
found to be inactive in WOC. This study focused on crystalline
manganese oxides of well-defined composition, but these rep-
resent only a small fraction of the known MnOx compounds.
Herein, we therefore extend the investigation to less-ordered
materials and present a systematic WOC analysis for twelve dif-
ferent MnOx oxides in which crystalline and amorphous oxides
of very different structures were screened by a common exper-
imental procedure. Surprisingly, such a screening is so far lack-
ing in the otherwise very active field of WOC by MnOx.
Therefore, we synthesised a representative variety of MnOx
compounds from the different structural types to probe the
role of the type of manganese oxide in water oxidation. Suita-
ble synthetic routes to the different phases were identified,
and the different MnOx products were first carefully character-
ised by a variety of analytical methods. Ion chromatography
(IC) in combination with redox titration was used to determine
the ion contents and average Mn oxidation states. From these
values, the chemical composition of the oxides could be calcu-
lated. XRD and IR spectroscopy provided information about
the atomic structure, crystallinity and water content of the
samples. Surface areas as well as particle shapes and sizes
were probed by nitrogen sorption and SEM. In combination
with WOC runs using the well-established chemical Ce4+
system, we are thus able to present a very detailed dataset on
For a short overview of common manganese oxide struc-
tures, some important manganese oxide phases are briefly de-
scribed in the following. For more detailed information, see
review articles on this topic.[14,24]
Manganese oxides are often categorised into three different
structure types, all of which contain [MnO6] units as basic
Chem. Eur. J. 2015, 21, 14958 – 14968
www.chemeurj.org
14959
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
probably the most diverse series
of MnOx water oxidation cata-
lysts screened in a single study
so far.
Table 1. Synthetic routes for the preparations of the different manganese oxides investigated in this study.
The chemical formulae given below the mineral names are generalised according to ref. [14]. For precise
values of the products, see Table 2.
Manganese oxide
bixbyite (Mn2O3)
Synthesis
Ref.
[27]
Results and Discussion
Synthesis of manganese oxides
hausmannite (Mn3O4)
[27]
[28]
The three classes of manganese
oxides are by now well de-
scribed, and many synthetic pro-
cedures leading to the various
oxides of each type have been
reported. Since the main motiva-
tion of our work was the identifi-
cation of catalysts for possible
large-scale applications, we se-
lected established MnOx synthe-
sis routes for this study that
yield the desired products from
low-cost precursors in simple
procedures. If precipitation reac-
tions led to the desired product,
we chose routes that involve
only aqueous media, because
water is the solvent and sub-
strate of WOC, and we thus
hoped to obtain water-stable
MnOx samples.
LiMn2O4
marokite (CaMn2O4)
[17]
pyrolusite (MnO2)
ramsdellite (MnO2)
[29]
[29]
cryptomelane [Kx(Mn3+,Mn4+)8O16]
[30,31]
[20,32]
Table 1 gives an overview of
the used synthetic routes. Inter-
estingly, one finds some general
differences in the reaction condi-
tions leading to the different
oxide types. Nearly all prepara-
tions start from MnII precursors,
but those of 3D networks com-
monly involve a final high-tem-
perature calcination step, tunnel-
led structures are mainly the
products of precipitation reac-
tions at elevated temperatures
and pressures (hydrothermal
synthesis) and layered oxides are
usually synthesised by MnII/MnVII
comproportionation reactions in
water at room temperature.
hollandite (MnO2·nH2O)
todorokite [(Ca,Na,K)x(Mn3+,Mn4+)6O12·nH2O]
[15]
Ca-birnessite [(Ca,Na)x(Mn3+,Mn4+)7O14·nH2O]
vernadite (MnO2·nH2O)
[9]
[33]
Thus, the systematics of the MnOx solid-state structures
(Figure 1) are to some degree mirrored by the reaction condi-
tions under which these oxides are typically formed.[8]
of the actual MnOx phase can be obtained from XRD measure-
ments on these products (Figure 2 and the Supporting Infor-
mation, Figure S2). For crystalline oxides, a correlation to the
theoretical diffractograms is mostly straightforward and thus
the products of the syntheses of bixbyite, hausmannite,
LiMn2O4, marokite, pyrolusite, cryptomelane and hollandite as
well as the commercial sample of manganosite could all clearly
be identified as the correct oxide phases. On the other hand,
the XRD patterns of the products of the syntheses of tunnelled
XRD and IR spectroscopy
The synthetic procedures shown in Table 1 all yield powders
with the brown to black colours typical of manganese oxides
(see the Supporting Information, Figure S1). A first indication
Chem. Eur. J. 2015, 21, 14958 – 14968
www.chemeurj.org
14960
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 3. FTIR spectra for selected synthetic manganese oxides focussing on
the spectral regions of the water and MnÀO bond stretching frequencies.
Additional FTIR spectra are shown in the Supporting Information, Figure S2.
these regions.[34] This interpretation is also in good agreement
with our analyses of the chemical compositions of the oxides
(see below): the water-related IR features are absent in the
spectra of crystalline, 3D-network materials (only O2À, but no
H2O/OHÀ, in these structures; see Table 2), whereas we found
pronounced water/hydroxide bands for layered oxides and
compounds containing large tunnels. The intensities of these
signals are also useful to confirm the relative amount of water
in the materials calculated from chemical analyses (see below):
for example, the higher water content found for the synthes-
ised Ca-birnessite in comparison to vernadite is supported by
the observation of stronger water bands in the IR spectrum of
the former.
In all cases the very broad H2O/OHÀ bands indicate that the
water-related species randomly occupy a number of non-
equivalent sites in the structures of the layered and tunnelled
oxides.[34] Furthermore, the intensity maxima for the bands be-
tween 3000 and 3600 cmÀ1 are always clearly above 3250 cmÀ1
and thus higher than expected for H2O/OHÀ moieties that are
part of well-defined hydrogen-bonding networks.[35]
Figure 2. Powder XRD patterns for different synthetic manganese oxides. In
each case, the theoretical patterns are shown as black bar graphs. Stars
mark reflections originating from the aluminium sample holders. Additional
diffractograms can be found in the Supporting Information, Figure S2.
or layered MnOx are as expected rather featureless, because it
is known that such oxides show only very little long-range
order. Hence, it is not possible to assign the oxide phases of
the isolated materials for ramsdellite, todorokite, birnessite and
vernadite from the XRD data alone, and especially for these
compounds further analytical methods must be employed to
identify oxide type and composition.
ATR-IR spectra were recorded for all synthetic oxides
(Figure 3 and the Supporting Information, Figure S3) and show
bands in three ranges: 500–700 cmÀ1, around 1600 cmÀ1 and
around 3300 cmÀ1. For all Mn3+- and/or Mn4+-containing com-
pounds, the most intense signals are the MnÀO vibrations in
the range of 500–700 cmÀ1. These originate from deformations
of the [MnO6] units and clearly differ between the individual
MnOx.[34] Amorphous oxides such as birnessite and todorokite
show broad bands, but sharp and defined signals are detected
for crystalline materials such as hausmannite. The MnÀO
stretching frequency of the MnII oxide manganosite is about
300 cmÀ1[34] and was thus out of the range of our diamond
ATR-IR set-up (lower detection limit: 500 cmÀ1).
In summary, the XRD and IR datasets make it possible to
roughly group the synthetic manganese oxides of this study
into two groups: 1) crystalline “true oxides” (e.g.. bixbyite,
manganosite) with sharp XRD and IR signals, in which O2À
anions are the only oxygen-containing species and 2) MnOx
materials for which broad XRD signals and pronounced water
IR bands are detected (e.g., hollandite, birnessite), both of
which are typical features of less-ordered MnOx materials con-
taining a significant amount of randomly incorporated H2O/
OHÀ in addition to O2À.
The IR bands at higher wavenumbers (1500–1700 cmÀ1 and
3000–3600 cmÀ1) are associated with lattice water molecules or
hydroxide ions,[35] and it is well known that H2O/OHÀ between
layers or in tunnels of certain MnOx cause IR absorptions in
Chem. Eur. J. 2015, 21, 14958 – 14968
www.chemeurj.org
14961
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 2. Selected analytical data and WOC rates for the synthetic manganese oxides investigated in this study.
Mn oxide
Mn content
[wt%][a]
Av. Mn oxidation
state[a]
SBET
[m2 gÀ1
WOC rate
WOC rate
[mmolO m ]
À2 hÀ1
2
Chemical formula
[mmolO molÀM1n hÀ1
]
[b]
]
2
bixbyite
hausmannite 71.2 (72.0)
LiMn2O4
marokite
manganosite 80.1 (77.4)
pyrolusite
ramsdellite[c] 61.6 (63.2)
69.6 (69.6)
3.0 (3.0)
2.7 (2.7)
3.5 (3.5)
2.9 (3.0)
2.0 (2.0–2.3)
3.9 (4.0)
3.7 (4.0)
3.8 (3.0–4.0)
3.9
3.5 (3.0–4.0)
3.7 (3.7–4.0)
4.0 (3.8–4.0)
22 (24)
11 (0.5)
70 (14)
1 (2.6)
2 (1.1)
12 (10)
95
90
0
50
30
0
50 (50)
0 (0)
10 (40)
285 (110)
0 (0)
15 (20)
50
25
40
50
60
Mn2O3.00
Mn3O4.02
Li0.92Mn2O3.50
Ca0.48MnO1.93
61.1 (60.7)
52.2 (51.4)
MnO1.05
62.7 (63.2)
10
MnO1.95·0.1H2O
MnO1.85·0.3H2O
K0.07MnO1.94
MnO1.95·0.4H2O
K0.11Ca0.22MnO2.03·0.7H2O
Ca0.21MnO2.16·1.3H2O
Na0.15K0.51MnO2.25·0.9H2O
425
300
275
1375
1650
425
Cryptomelane 61.6 (43.5–63.2)
125
hollandite[d]
todorokite
birnessite
vernadite
58.7
80
250
240
160
48.7 (29.2–39.0)
46.0 (38.5–39.1)
41.8 (53.9–68.7)
20
[a] In parentheses: typical values of naturally occurring mineral samples;[14] [b] from N2 sorption isotherms (in parentheses: calculated values Scalcd from par-
ticle sizes and densities; see the Supporting Information for details); [c] SEM and XRD suggest that the synthesised material contains significant amounts
of birnessite impurities; [d] not a naturally occurring mineral.
Chemical composition
In summary, the XRD, IR and chemical analyses of crystalline
Using a previously established methodology, we next deter-
mined chemical formulae for the synthetic oxides by a combi-
nation of IC and redox titration (for details, see the Supporting
Information and ref. [9]). The results for the different synthetic
manganese oxides are listed in Table 2. A comparison between
theoretical and experimentally determined Mn contents and
oxidation states shows very good correlations for the crystal-
line 3D-network oxides bixbyite, hausmannite, LiMn2O4, maro-
kite and manganosite. In these cases the oxide phases indicat-
ed by XRD closely match the results of the chemical analysis.
Reasonable correlations between phase assignments by XRD
and chemical formulae were also found for pyrolusite and
ramsdellite, whose structures are characterised by small tun-
nels (types [1”1] and [2”1], respectively). These tunnels are
too small to accommodate additional cations, which could be
confirmed by IC, since only Mn2+ was found. On the other
hand, these materials show some disorder (e.g., see the broad
XRD peaks in Figure 2), and small amounts of water can appa-
rently enter the structures (see IR spectra, Figure 3). In conse-
quence, the manganese contents are in both cases slightly
lower than the theoretical values and in agreement with for-
mulae that contain 0.1–0.3 equivalents of H2O per Mn.
manganese oxides and MnOx materials with structures contain-
ing narrow tunnels is rather straightforward and it is possible
to confirm the isolation of the intended synthetic products
with great certainty. On the other hand, the detectable differ-
ences in composition between the highly disordered MnOx
compounds birnessite, todorokite and vernadite are small, and
thus an unambiguous assignment of the identities of the com-
pounds without more expensive analytical tools such as X-ray
absorption spectroscopy or X-ray photoelectron spectroscopy
is difficult.[8] However, for our set of samples a clear difference
between the three products is the average manganese oxida-
tion state, which is, as expected from the mineralogical data,[14]
found to be lowest for todorokite and highest for vernadite
with an intermediate value for the birnessite sample. As the ac-
curacy of this value obtained by our method is about Æ0.1,
the detected trend does not provide an ultimate proof, but at
least a very good indication, that the chosen synthetic proce-
dures indeed yielded the desired tunnelled or layered manga-
nese oxides.
Morphology
For the oxides with wider tunnel geometries, such as crypto-
melane ([2”2]) and todorokite ([3”3]), the cavities in the
structure are large enough to accommodate both water and
additional cations, especially M+/2+ ions of alkali and alkaline
earth metals, which are often present during syntheses in
aqueous solution. This results in a large variety of possible
cation and water contents, and thus complex chemical compo-
sitions are found. Often, the determined compositions differ
significantly from those of the corresponding geologically oc-
curring minerals. This compositional flexibility is known to be
even more pronounced for layered oxides such as birnessite
and vernadite (see above), and the data for the layered com-
pounds listed in Table 2 reflect this very well.
Nitrogen sorption data, analysed by the BET method, were
used to estimate the surface areas SBET of the oxides. For the
studied MnOx series the measured values for SBET vary over
a large range of 1–250 m2 gÀ1 (Table 2), and SBET is a parameter
that is greatly influenced by the oxide type. In general, crystal-
line materials show (as expected) the smallest surface areas,
and much larger SBET values are found for tunnelled and espe-
cially layered oxides. To some degree surface areas correlate
with crystallinity, but it is remarkable that even a rather well-or-
dered material such as hollandite (see XRD pattern in Figure 2)
can show a high surface area of 80 m2 gÀ1.
The differences in surface areas might be explained by the
nanostructured surfaces of the synthetic manganese oxides re-
vealed by SEM (Figure 4). Again, the three oxide types also
Chem. Eur. J. 2015, 21, 14958 – 14968
www.chemeurj.org
14962
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 4. SEM images of different manganese oxides at the same magnification: a) bixbyite; b) hausmannite; c) LiMn2O4; d) marokite; e) manganosite; f) pyro-
lusite; g) ramsdellite; h) cryptomelane; i) cation-free hollandite; j) todorokite; k) birnessite; l) vernadite.
differ in this material property: tunnelled oxides can often be
recognised by needle-like appearances (Figure 4 f–h), and be-
cause these needles are only few nanometres thick, large sur-
face areas result (e.g., for cryptomelane). In contrast, crystalline
materials such as bixbyite and hausmannite (Figure 4a,b) con-
sist of round, much larger particles with sizes of 0.5–10 mm.
These are themselves sometimes built up from smaller spheri-
cal domains of 50–100 nm in diameter, as illustrated by the
SEM image of LiMn2O4 (Figure 4c).
ticles, since we found the broad peaks of birnessite in the XRD
pattern and also detected a significantly lower average Mn oxi-
dation state than expected from mineralogical data. However,
additional cations (e.g., NH4+; see synthesis protocol) could
not be detected, and therefore our impression from the SEM
images that the material mainly consists of ramsdellite was
confirmed.
From the presented analytical data, we conclude that com-
pounds from all three MnOx structure families were successful-
ly synthesised. Manganese oxides of 3D-network structures
show well-defined XRD patterns, chemical compositions very
close to the theoretical values and consist of rather large parti-
cles with low surface areas. In contrast, tunnelled and layered
oxides are generally much less ordered, more diverse in their
chemical compositions and contain smaller particles with quite
large surface areas of up to 250 m2 gÀ1. We therefore find vari-
ous aspects of diversity for this series of manganese oxides:
chemical composition, Mn oxidation state, arrangement of the
ions at the atomic level and different particle morphologies.
The most irregular particle morphologies are found for the
amorphous oxides todorokite, birnessite and vernadite (Fig-
ure 4j–l). All are characterised by rather large grains (>5 mm)
that are aggregates of very fine (d<10 nm) needles (todoro-
kite) or irregular particles (birnessite, vernadite). Overall, non-
compact, sponge-like structures are the result, especially well
illustrated by the SEM image of birnessite (Figure 4k).
The highly textured particle morphologies detected by SEM
often correlate well with the SBET values. Additionally, if the N2
sorption data are analysed with regard to the porosity of the
samples, one finds that most of the tunnelled and layered
oxides are mesoporous materials with median pore diameters
of about 15 nm (see the Supporting Information, Figure S4 and
Table S2). The quite high surface areas of more than 100 m2 gÀ1
found especially for some of the samples synthesised by pre-
cipitation reactions can therefore be explained by the fact that
the materials have pronounced surface textures and significant
mesoporosity.
Water oxidation catalysis
The WOC activities of the synthetic manganese oxides were
determined with Ce4+ as chemical oxidising agent. This reac-
tion has been commonly used for this purpose in the
past[9,17,36] and is a good WOC screening method because:
a) Ce4+ is a strong (but not extreme) single-electron oxidant
with an oxidation potential of about+1.6 V versus NHE;[37]
2) 18O labelling has shown that the oxygen formed in reactions
of Ce4+ with MnOx in water originates from the solvent and
thus “true” water oxidation occurs[38] and 3) solutions of Ce4+
are stable for hours, and thus extended catalysis runs can be
For the product of the synthesis of ramsdellite, we detected
a mixture of two crystal forms by SEM: the fine needles typical
of a tunnelled oxide, but also some sponge-like objects, which
constitute about 10% of the material (Supporting Information,
Figure S5). We suspect that these impurities are birnessite par-
Chem. Eur. J. 2015, 21, 14958 – 14968
www.chemeurj.org
14963
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
carried out.[9] As a drawback, Ce4+ is only soluble under acidic
conditions, and thus catalysis experiments are usually carried
out at pHꢀ2.[8,37]
during the monitored first hour of catalysis (see the Support-
ing Information). As a result, [Ce4+] is markedly decreased and
significant [Ce3+] builds up, both possibly resulting in the slow-
ing down of the rate of O2 evolution after about 30 min that is
observed in experiments with high WOC rates (Figure 5).
There are different options to analyse the data for compari-
son of WOC rates. The most common methods are calculations
of the catalytic rate per unit mass, per mole of manganese
ions and per unit surface area, and all of them have advantag-
es and disadvantages. Because manganese centres are essen-
tial for WOC (at least in the absence of other potentially active
metals such as Co, Ru and Ir, as is the case here), we (and
others) have usually decided to plot the data per mole of
Mn.[9,19,21,36] This was also done here to generate the plots
shown in Figure 5 or to derive the averaged rates for the first
hour of reaction given in Figure 6 (top) and Table 2. In addi-
tion, comparisons per unit mass or per unit surface area are
also shown (Figure 6 (bottom), Table 2 and the Supporting In-
formation, Figure S6), and especially the surface-normalised
values are discussed below.
As in our previous experiments,[9] the synthetic oxides were
suspended in air-saturated aqueous solutions of Ce4+
([MnOx]=1 mgmLÀ1, [Ce4+]=250 mm) in capped vials and
kept at 408C. Over 1 h, gas samples were taken from the head-
space (air) every ten minutes and analysed by gas chromatog-
raphy for their oxygen-to-nitrogen ratios. The amount of
evolved oxygen can then be calculated from the increasing
percentage of O2 in the headspace above the catalytic mixture.
The slightly elevated temperature was chosen to make com-
parisons with our earlier studies on water oxidation catalysis
by manganese oxide possible, whereby the higher-than-ambi-
ent temperatures were a requirement of the previously used
detection system.[9,10] For further information on the catalysis
experiments, see the Supporting Information and ref. [9].
The examples of typical oxygen-evolution traces in Figure 5
illustrate that, depending on the oxide, no (e.g., manganosite),
Figure 5. Exemplary oxygen evolution traces for reactions of Ce4+ with
some of the synthetic manganese oxides prepared for this study. The addi-
tional O2 accumulated in the headspace above the suspensions is plotted.
Reaction conditions: 5 mL aqueous suspension, [MnOx]=1 mgmLÀ1
[Ce4+]=250mm, T=408C.
,
slow (e.g., marokite) or much faster (e.g., todorokite) WOC is
observed. The cases in which significant amounts of product
are detected generally show no extended lag phase at the be-
ginning of the experiment and some O2 is detectable already
at the first time of sampling (t=10 min). Furthermore, the
traces for oxides with slow to intermediate rates show a con-
stant WOC rate for the first hour, whereas fast catalysis (todor-
okite, birnessite) tends to slow down slightly with time. For bir-
nessites, we have observed this behaviour before and con-
firmed that the rate decrease is not a result of MnOx corro-
sion.[9,10] Instead, we suspect that a combination of progressive
Ce4+ consumption and the back-reaction of Ce3+ with O2
might be reasons for this observation. As an example, for the
fastest reaction rate found here (Ca-birnessite), one can calcu-
late that about one quarter of the Ce4+ ions are consumed
Figure 6. Comparison of catalytic water oxidation rates for the different
oxides with Ce4+ as oxidant. Top: Data normalised per mole of Mn. Bottom:
Rates per unit surface area determined by N2 physisorption (front row,
black) or calculated from particle sizes (back row, grey). Reaction conditions
as in Figure 5. The error bars indicate the estimated 10% error of the abso-
lute value. a) The values for potassium birnessite were taken from ref. [10].
Chem. Eur. J. 2015, 21, 14958 – 14968
www.chemeurj.org
14964
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
As the main result of the catalysis data normalised per mole
of Mn (Figure 6, top), we find two general trends for WOC by
the different MnOx: 1) most crystalline, 3D-network oxides
show little or no activity in WOC and 2) for the groups of tun-
nelled and layered oxides, the best rates are clearly detected
for todorokite and Ca-birnessite. In their respective groups of
oxides, these are the two least-ordered materials. They show
the highest surface areas/porosities and have accessible inter-
nal volumes due to their large tunnel diameters and interlayer
spaces, respectively, and overall all these factors might explain
the good catalytic performances of todorokite and Ca-birnes-
site.
conclude that, in addition to Mn3+/4+ and O2À, additional ions
and/or water are also beneficial for obtaining a good Mn-
based WOC material. However, we note that preliminary ex-
periments indicated that the situation for mixed oxides ap-
pears to be very dynamic. As we found that cerium ions are
able to replace some of the alkali/alkaline earth cations over
time under “cerium conditions”. From our initial data, this ap-
pears to be a rather complex process depending on oxide
type, [Ce4+] and reaction time and thus requires further inves-
tigations beyond the scope of the present study. 3) For tunnel-
led oxides, the tunnel size influences catalytic activity, since
larger diameters ([2”2] and especially [3”3]) seem to be ad-
vantageous, even though the data for ramsdellite must be
treated with caution, because this material contained birnes-
site impurities (see above). We suspect that large tunnel diam-
eters make more active sites accessible for water oxidation,
either because Ce4+ itself enters the structure or because in-
jected holes move through the material and water molecules
inside the tunnels are then oxidised.[5,9,23,25] 4) WOC clearly de-
pends on the crystallinity of the samples, since most of the
best-performing materials are amorphous MnOx with high spe-
cific surface areas. On the other hand, some crystalline materi-
als such as marokite and bixbyite can also show impressive
catalytic rates (at least per unit SBET) if they fulfil some of the
other advantageous criteria, such as incorporated Ca2+ and/or
an intermediate Mn oxidation state of +III.
The two above-mentioned general trends are most obvious
when the data are normalised per manganese centre or per
unit catalyst mass. If rates per unit surface area are compared,
birnessite and todorokite still show very good rates, but three
other oxides now also score very well: bixbyite, marokite and
ramsdellite (Figure 6, bottom). These three thus also belong to
the MnOx showing good catalyst performance, and in the case
of Mn2O3 this is in agreement with the data of Robinson
et al.[20] However, especially the first two of these materials
belong to the group of crystalline oxides with small surface
areas (SBET =22 and 1 m2 gÀ1, respectively), and the accuracy of
N2 sorption measurements for such small values is low. To vali-
date the SBET values, we therefore also approximated specific
surface areas by calculations in which theoretical densities
from the crystal structures and (idealised) particle shapes from
the SEM micrographs were used (see the Supporting Informa-
tion for details). These calculated values Scalcd are also listed in
Table 2, and for some materials, Scalcd and SBET match rather
well. For others, large differences are found, and as a result,
the catalytic rates per surface areas can be very different de-
pending on whether SBET or Scalcd is used (Table 2 and Figure 6,
bottom). Overall, this indicates that an analysis of catalytic
rates per unit SBET (though definitely important) is not very pre-
cise for materials of low surface areas.
Conclusion
A series of manganese oxides has been investigated for use in
WOC. The oxides can be classified according to their structural
properties as 3D-network, tunnelled or layered materials. In ad-
dition to the arrangement of the ions at the atomic level, the
three oxide families also differ when synthesis routes, crystal-
linities and particle morphologies are compared. For the rates
of WOC with Ce4+ as oxidant, we found a general reactivity
trend of 3D networks<tunnelled structuresꢀlayered struc-
tures. Interestingly, this is in agreement with a related (but
very limited) first study by us on electrocatalytic water oxida-
tion with three different MnOx, for which a ranking Ca-birnes-
site>Mn2O3 >MnO2 was found when anodes coated by these
oxides were used for water electrolysis at pH 7.[39] Given the
large differences in catalytic performance found here, we will
now carry out a similar screening to assess the suitability of
the different MnOx for water-oxidation electrocatalysis. Because
electrochemical water-oxidation rates strongly depend on pa-
rameters such as pH and oxidation potential, but also oxide
conductivity,[8] a comparison of the Ce4+ and electrolysis data-
sets will likely offer interesting insights into the catalytic pro-
cess.
What are the general properties emerging from these data
that good MnOx water oxidation catalysts have in common?
We see at least four key points: 1) The oxidation state of man-
ganese should allow changes between the critical oxidation
states[26] of +III and +IV while catalysis takes place. Hence,
oxides that contain only Mn2+ (manganosite) or only Mn4+
(pyrolusite, hollandite) often perform poorly. In contrast, bixby-
ite, ramsdellite, cryptomelane, todorokite etc. also contain
Mn3+ ions and therefore likely more active sites for water oxi-
dation. 2) Incorporated ions and/or water can be beneficial for
catalytic activity. Former studies revealed an important role for
secondary cations in birnessites, in which (in analogy to PSII)
Ca2+ leads to best catalytic results, followed by Sr2+ and K+
(see difference between Ca and K birnessites in Figure 6, top,
and the Supporting Information, Figure S6).[10,13] The data pre-
sented here are in agreement with these previous conclusions,
since the two fastest catalysts (per mole Mn or gramme) are
again the two calcium-containing compounds of the series,
Ca-birnessite and todorokite. More generally, it is striking that
Overall, we firstly conclude that good WOC activity is found
for amorphous materials containing highly flexible MnÀO link-
ages. Secondly, the structures should offer room for additional
cations and/or water, since water oxidation most probably
takes place not only on the particle surfaces, but also in tun-
nels or between [MnO6] layers. Finally, two crystalline oxides
(bixbyite and marokite) also showed surprisingly high catalytic
all oxides reaching rates greater than 200 mmolO molÀM1n hÀ1
2
contain potassium or water in their structures, and thus we
Chem. Eur. J. 2015, 21, 14958 – 14968
www.chemeurj.org
14965
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Determination of the average manganese oxidation state by
redox titration: The method followed an established procedure
for the quantification of MnO2.[41] Carefully weighed samples of the
oxides (100–150 mg) were mixed with a solution of sodium oxalate
(10 mL, 0.2m) and sulfuric acid (10 mL, 0.5m). The solution was
stirred at 608C until all oxide had dissolved. In this step manganese
is quantitatively reduced to its Mn2+ form. In the following, the
excess of oxalate was determined by titration of the remaining ox-
alate with a solution of KMnO4. In combination with the manga-
nese content of the material (determined by IC; see above), the
average oxidation state of the manganese centres could then be
calculated.
activities per unit surface area, and future studies should at-
tempt to synthesise such oxides in nanostructured forms. If
WOC activities per unit area remain unaltered, these crystalline
materials may prove to be an alternative to the MnOx water
oxidation catalysts with layered or wide-tunnelled structures,
which are otherwise the clear winners of the above-presented
screening.
Experimental Section
General considerations
Oxygen detection by GC: To determine the concentration of
oxygen in the headspace of the reaction vessels, gas chromato-
grams were recorded. We used a PerkinElmer Clarus 480 gas chro-
matograph with a 12 ft.”1/8 in. 5 Š Molsieve column (Restek). The
carrier gas was helium, the GC oven temperature was set to 708C
and O2 and N2 were quantified by a thermal conductivity detector
(TCD). Ambient air served as calibration gas.
All reagents and solvents were purchased from commercial sources
and were used without further purification. Deionised water was
used for all experiments and, unless otherwise stated, all syntheses
described below were carried out in aqueous solutions. The
sample of manganosite (MnO) was obtained from ChemPur. For
centrifugation we used a Universal 320 centrifuge from Hettich
Zentrifugen. If not stated otherwise, centrifugation was carried out
for 2 min at 3000 rpm. For details of analytical procedures and cat-
alysis runs to probe WOC activity, see the Supporting Information
and ref. [9].
Determination of WOC rates for experiments with CeIV: The
method followed a previously developed procedure.[9] Precisely
weighed oxide samples (ca. 5 mg) and (NH4)2Ce(NO3)6 (685 mg,
1.25 mmol) were filled into 20 mL septum vials. After addition of
5 mL of air-saturated water, the vials were capped immediately
with gas-tight septa and sonicated for 20 s. The reaction containers
were kept at 408C in a water bath, and headspace gas samples
(100 mL) were injected into the gas chromatograph by hand with
a gas-tight syringe (Hamilton). Six injections at 10 min intervals
were carried out. The amount of oxygen evolved was then calcu-
lated for each headspace extract from the detected O2/N2 signal
ratios. To do so, the amount of oxygen from air (corresponding to
the detected nitrogen peak) was calculated and subtracted to
leave the excess of O2 generated by the water-oxidation reaction.
We estimate the error margin for the rate determination to be in
the range of Æ10%, as is indicated by error bars in Figure 6. For
more information, see the supporting information of Frey et al.[9]
Experimental details and analytical procedures
Infrared spectroscopy: IR spectra of the neat solid compounds
were measured with a Thermo Scientific Nicolet iS10 FTIR spec-
trometer equipped with a diamond ATR unit.
Powder XRD: Powder XRD patterns were recorded in reflection ge-
ometry with a Seifert 3003 TT instrument and CuKa radiation. Oxide
powders were mounted on aluminium frames exposing an area of
about 4 cm2 of the MnOx powders to the X-ray beam, which hit
the sample at an incident angle of 18.
BET analysis: N2 physisorption isotherms were measured with
a Thermo Porotec Sorptomatic 1990 instrument and analysed ac-
cording to the BET theory to determine SBET. Pore volumes and di-
ameters were calculated with the data analysis software Sorpto-
matic provided by Thermo and are listed in the Supporting Infor-
mation (Table S2) together with general porosity types according
to ref. [40]. Prior to analysis, oxide samples were heated at 1508C
for 5 h and about 10À5 bar. For water-containing samples, post-
measurement IR spectra indicated that this treatment does not
result in significant loss of intercalated H2O molecules. As N2 physi-
sorption does not result in very accurate data for SBET <20 m2 gÀ1
we additionally approximated surface areas for samples of low SBET
from the particle sizes detectable by SEM (see the Supporting In-
formation).
Syntheses of manganese oxides
Bixbyite: Analogous to the literature procedure,[27] MnCO3 (4.19 g)
was heated to 6008C for 24 h in air. Yield: 98%.
Hausmannite: The synthesis of hausmannite was adapted from
ref. [27]. MnCO3 (6.73 g) was heated to 10008C for 24 h in air.
Yield: 89%.
LiMn2O4: LiMn2O4 was prepared via a sol–gel route according to
the procedure established by Vivekanandhan et al.[28] A solution of
citric acid (28.8 g, 150 mmol) and urea (9.01 g, 150 mmol) in water
(100 mL) was added dropwise to a solution of LiNO3 (1.72 g,
2.5 mmol) and Mn(OAc)2·4H2O (12.3 g, 5.00 mmol) in water
(100 mL). Concentrated HNO3 (20 mL) was then added slowly with
stirring. The water was evaporated completely with a rotary evapo-
rator and the yellow precipitate was heated to 1708C for 12 h. Sub-
sequently, the brown, sponge-like solid was calcined twice at
3508C for 12 h. The product was ground with a pestle and mortar
after each calcination step. Yield: 77%.
Marokite: Marokite has been previously synthesised by us.[17]
Ca(NO3)2·4H2O (0.473 g, 2.00 mmol) and Mn(NO3)2·4H2O (0.703 g,
2.80 mmol) were dissolved in water (5 mL). A solution of KMnO4
(0.190 g, 1.20 mmol) and KOH (8.40 g, 150 mmol) in water (10 mL)
was added dropwise with vigorous stirring within 15 min. The sus-
pension was allowed to react for another 15 min. The dark-brown
precipitate was separated by centrifugation and washed with
water (5”200 mL) and centrifuged again. The crude product was
,
SEM and EDX measurements: SEM imaging and energy-dispersive
X-ray (EDX) analysis of the oxides were performed with a LEO 1525
scanning electron microscope at 3 kV accelerating voltage.
Ion chromatography: Contents of manganese, calcium, potassium,
sodium and lithium ions were determined with a Metrohm 882
Compact IC plus ion chromatograph equipped with a Metrosep C4
150/4 column and a conductivity detector. A solution of dipicolinic
acid (0.75 mm) and nitric acid (2 mm) served as IC eluent, and the
chromatograph was calibrated with AAS standard solutions (Roth).
Prior to analysis, the oxide samples (ca. 10 mg) were treated with
1 mL of concentrated HNO3/30% H2O2 (1/10), which completely
dissolved the oxide and converted all manganese to its Mn2+ form.
The solutions were then diluted to 250 mL, and 20 mL samples
were injected into the IC system.
Chem. Eur. J. 2015, 21, 14958 – 14968
www.chemeurj.org
14966
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
then dried at 608C for 6 h and calcined at 10008C for 10 h. Yield:
87%.
The precipitate was obtained by centrifugation. The brown prod-
uct was dried at room temperature. Yield: 98%.
Pyrolusite: The synthesis was adapted from Ghodbane et al.[29]
First, a lithium-doped birnessite was synthesised: Mn(OAc)2·4H2O
(3.92 g, 16.0 mmol) in water (30 mL) was added with stirring to
a solution of KOH (10.1 g) in water (30 mL). A solution of KMnO4
(0.949 g, 6.00 mmol) in water (100 mL) was added dropwise with
vigorous stirring to form a brown suspension. The mixture was al-
lowed to age for 2 h with gentle stirring, after which the crude
product was obtained by centrifugation. The supernatant was dis-
carded and the precipitate was suspended in water (200 mL) and
again centrifuged. This washing procedure was carried out seven
times in total. The product was then suspended in an aqueous so-
lution of LiCl (1 molLÀ1, 200 mL) and the suspension was stirred for
24 h. The suspension was centrifuged and the centrifugate washed
with water (200 mL), after which it was separated by centrifugation
again. This cation exchange was repeated once more. The brown
precipitate was dried at 608C for 24 h. Part of this lithium-doped
birnessite (ca. 400 mg) was suspended in a solution of H2SO4
(0.5 molLÀ1) und LiCl (1 molLÀ1) in water (10 mL). The suspension
was transferred to a Teflon-lined autoclave (15 mL) and heated to
1508C for 48 h. The dark-brown precipitate was separated by cen-
trifugation, washed with water (3”50 mL), and centrifuged again.
The dark brown product was then dried at 608C for 48 h. Yield:
64%.
Birnessite: For the synthesis of a Ca-birnessite a well-established
route was used that was developed by Suib et al.[15,42] and modi-
fied by us.[9,10] Three solutions were prepared: solution A: KOH
(14.4 g, 250 mmol) in H2O (30 mL); solution B: Mn(OAc)2·4H2O
(3.92 g, 16.0 mmol) and Ca(OAc)2·H2O (0.846 g, 4.80 mmol) in H2O
(30 mL); solution C: KMnO4 (0.948 g, 6.00 mmol) in H2O (100 mL).
First solution B was added dropwise over 10 min with stirring to
solution A to give an off-white to light brown suspension. Then,
solution C was added dropwise to the vigorously stirred mixture in
30 min. After gently stirring the resulting dark brown to black sus-
pension for 2 h at room temperature, the crude product was ob-
tained by centrifugation. The supernatant was discarded, and the
precipitate was suspended in H2O and again centrifuged. This
washing procedure was carried out seven times in total. The pre-
cipitate was dried at 608C for 48 h. Yield: 95%.
Vernadite: This synthesis was adapted from Villalobos et al.[33] As
described in the synthesis of birnessite three solutions were pre-
pared: solution A: NaOH (2.78 g, 69.5 mmol) in H2O (130 mL); solu-
tion B: KMnO4 (4.00 g, 25.3 mmol) in H2O (180 mL); solution C:
MnCl2·4H2O (7.52 g, 3.80 mmol) in H2O (130 mL). First, solution B
was added over 5 min to solution A with stirring. Subsequently, sol-
ution C was added to the mixture over 20 min with vigorous stir-
ring. The brown suspension was allowed to age with gentle stirring
for 30 min, after which the precipitate was allowed to settle down
for 2 h. The light-brown supernatant was discarded. The obtained
crude product was re-suspended in water (90 mL) and centrifuged.
The obtained brown product was dried at room temperature for
72 h. Yield: 91%.
Ramsdellite: This synthesis was adapted from Ghodbane et al.[29]
MnSO4·4H2O (0.423 g, 2.50 mmol) and (NH4)2S2O8 (0.571 g,
2.50 mmol) were dissolved in water (8 mL) and the solution was
transferred to a Teflon-lined autoclave (15 mL). The solution was
heated to 858C for 12 h. The dark-brown crude product was sepa-
rated by centrifugation, washed with water (50 mL) and centri-
fuged again. This washing procedure was repeated twice. The
product was then dried at 708C for 72 h. Yield: 66%.
Acknowledgements
Cryptomelane: This oxide was prepared by following a procedure
developed by Luo et al.[30,31] KMnO4 (6.01 g, 38.0 mmol) in water
(30 mL) was added dropwise to a solution of MnSO4·H2O (10.1 g,
60.0 mmol) and HNO3 (3 mL) in water (100 mL) at 110 8C. The
brown suspension was heated to reflux for 24 h. The crude brown
product was obtained by centrifugation. The colourless superna-
tant was discarded and the precipitate was suspended in water
(200 mL) and again centrifuged. This washing procedure was car-
ried out five times in total. The greyish black precipitate was dried
at 1208C for 12 h and ground into fine powder with a pestle and
mortar. Yield: 98%.
At the ALU Freiburg, we would like to thank Katrin Kaspar for
obtaining SEM images, Sabine Zuelsdorf and Simon Rudolf for
assistance with the chemical analyses, Andreas Warmbold for
BET measurements and Robin White for advice on the analysis
of sorption isotherms. The Freiburger Materialforschungszen-
trum (FMF) granted access to its XRD instrument. This work
was financially supported by the Deutsche Forschungsgemein-
schaft (DFG projects KU 2885/1-1 and KU 2885/2-1, the latter
part of the priority program SPP 1613).
Cation-free hollandite: This synthesis was developed by Johnson
et al.[32] and modified by Robinson et al.[20] Mn2O3 (bixbyite, 2.05 g,
13.0 mmol; see above) was heated to 7008C for 12 h to remove
impurities. The residue was added to an aqueous solution of H2SO4
(4 molLÀ1) and stirred at 808C for 6 h. After the suspension was al-
lowed to cool the crude product was obtained by centrifugation.
The supernatant was discarded and the precipitate was suspended
in water (200 mL) and centrifuged. This washing procedure was
carried out three times in total. The brown precipitate was dried at
908C for 16 h. Yield: 93%.
Keywords: heterogeneous catalysis · manganese · oxidation ·
oxides · water splitting
[1] a) REN21 Renewable Energy Policy Network, Renewables 2014 Global
Status Report; b) A. Thapper, S. Styring, G. Saracco, A. W. Rutherford, B.
Robert, A. Magnuson, W. Lubitz, A. Llobet, P. Kurz, A. Holzwarth, S.
Fiechter, H. de Groot, S. Campagna, A. Braun, H. Bercegol, V. Artero,
Green 2013, 3, 43–57.
[2] J. Barber, Chem. Soc. Rev. 2009, 38, 185–196.
[3] M. Suga, F. Akita, K. Hirata, G. Ueno, H. Murakami, Y. Nakajima, T. Shimi-
zu, K. Yamashita, M. Yamamoto, H. Ago, J.-R. Shen, Nature 2014, 517,
99–103.
[4] a) N. Cox, D. A. Pantazis, F. Neese, W. Lubitz, Acc. Chem. Res. 2013, 46,
1588–1596; b) P. E. Siegbahn, Acc. Chem. Res. 2009, 42, 1871–1880.
[5] H. Dau, C. Limberg, T. Reier, M. Risch, S. Roggan, P. Strasser, ChemCat-
Chem 2010, 2, 724–761.
[6] T. S. Glikman, I. S. Shchegoleva, Kinet. Katal. 1968, 9, 461–462.
[7] V. Y. Shafirovich, N. K. Khannanov, A. E. Shilov, J. Inorg. Biochem. 1981,
15, 113–129.
Todorokite: The preparation was carried out by following a publica-
tion by Luo et al.[15] A solution of Mn(OAc)2·4H2O (0.784 mg,
3.20 mmol) and Ca(OAc)2·H2O (0.163 mg, 0.925 mmol) in water
(5 mL) was added dropwise to
a solution of KOH (2.90 g,
51.7 mmol) in water (6 mL). KMnO4 (0.190 mg, 1.20 mmol) was
added slowly with vigorous stirring. The brown suspension was al-
lowed to age for 1 h. The crude product was obtained by centrifu-
gation, re-suspended in water (10 mL) and filled into a Teflon-lined
autoclave (15 mL). The mixture was then heated to 1508C for 48 h.
Chem. Eur. J. 2015, 21, 14958 – 14968
www.chemeurj.org
14967
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[8] P. Kurz, Topics Curr. Chem. 2015, DOI: 10.1007/128_2015_634.
[9] C. E. Frey, M. Wiechen, P. Kurz, Dalton Trans. 2014, 43, 4370.
[10] M. Wiechen, I. Zaharieva, H. Dau, P. Kurz, Chem. Sci. 2012, 3, 2330–
2339.
[28] S. Vivekanandhan, M. Venkateswarlu, N. Satyanarayana, J. Alloys Compd.
2007, 441, 284–290.
[29] O. Ghodbane, J.-L. Pascal, B. Fraisse, F. Favier, ACS Appl. Mater. Interfaces
2010, 2, 3493–3505.
[11] F. Jiao, H. Frei, Chem. Commun. 2010, 46, 2920–2922.
[12] A. Iyer, J. Del-Pilar, C. K. King’ondu, E. Kissel, H. F. Garces, H. Huang,
A. M. El-Sawy, P. K. Dutta, S. L. Suib, J. Phys. Chem. C 2012, 116, 6474–
6483.
[30] S. Luo, L. Duan, B. Sun, M. Wei, X. Li, A. Xu, Appl. Catal. B 2015, 164,
92–99.
[31] A. Iyer, H. Galindo, S. Sithambaram, C. King’ondu, C.-H. Chen, S. L. Suib,
Appl. Catal. A 2010, 375, 295–302.
[13] M. M. Najafpour, D. J. Sedigh, B. Pashaei, S. Nayeri, New J. Chem. 2013,
37, 2448.
[14] J. E. Post, Proc. Natl. Acad. Sci. USA 1999, 96, 3447–3454.
[15] J. Luo, Q. H. Zhang, A. M. Huang, O. Giraldo, S. L. Suib, Inorg. Chem.
1999, 38, 6106–6113.
[32] C. S. Johnson, D. W. Dees, M. F. Mansuetto, M. M. Thackeray, D. R. Viss-
ers, D. Argyriou, C.-K. Loong, L. Christensen, J. Power Sources 1997, 68,
570–577.
[33] M. Villalobos, B. Toner, J. Bargar, G. Sposito, Geochim. Cosmochim. Acta
2003, 67, 2649–2662.
[16] I. Zaharieva, M. M. Najafpour, M. Wiechen, M. Haumann, P. Kurz, H. Dau,
Energy Environ. Sci. 2011, 4, 2400–2408.
[17] M. M. Najafpour, T. Ehrenberg, M. Wiechen, P. Kurz, Angew. Chem. Int.
Ed. 2010, 49, 2233–2237; Angew. Chem. 2010, 122, 2281–2285.
[18] A. Harriman, I. J. Pickering, J. M. Thomas, P. A. Christensen, J. Chem. Soc.
Faraday Trans. 1 1988, 84, 2795–2806.
[19] V. Boppana, B. Ram, F. Jiao, Chem. Commun. 2011, 47, 8973–8975.
[20] D. M. Robinson, Y. B. Go, M. Mui, G. Gardner, Z. Zhang, D. Mastrogiovan-
ni, E. Garfunkel, J. Li, M. Greenblatt, G. C. Dismukes, J. Am. Chem. Soc.
2013, 135, 3494–3501.
[21] M. Fekete, R. K. Hocking, S. L. Y. Chang, C. Italiano, A. F. Patti, F. Arena, L.
Spiccia, Energy Environ. Sci. 2013, 6, 2222.
[22] Y. Okuno, O. Yonemitsu, Y. Chiba, Chem. Lett. 1983, 815–818.
[23] Y. Gorlin, B. Lassalle-Kaiser, J. D. Benck, S. Gul, S. M. Webb, V. K. Yachan-
dra, J. Yano, T. F. Jaramillo, J. Am. Chem. Soc. 2013, 135, 8525–8534.
[24] B. M. Tebo, J. R. Bargar, B. G. Clement, G. J. Dick, K. J. Murray, D. Parker,
R. Verity, S. M. Webb, Annu. Rev. Earth Planet. Sci. 2004, 32, 287–328.
[25] G. Elmaci, C. E. Frey, P. Kurz, B. Zümreog˘lu-Karan, Inorg. Chem. 2015, 54,
2734–2741.
[34] R. M. Potter, G. R. Rossman, Am. Mineral. 1979, 64, 1199–1218.
[35] K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination
Compounds, Wiley, Hoboken, NJ, 2009.
[36] M. M. Najafpour, B. Pashaei, S. Nayeri, Dalton Trans. 2012, 41, 4799–
4805.
[37] A. F. Holleman, E. Wiberg, Lehrbuch der Anorganischen Chemie, De
Gruyter, Berlin, 2007.
[38] D. Shevela, S. Koroidov, M. M. Najafpour, J. Messinger, P. Kurz, Chem.
Eur. J. 2011, 17, 5415–5423.
[39] S. Y. Lee, D. Gonzµlez-Flores, J. Ohms, T. Trost, H. Dau, I. Zaharieva, P.
Kurz, ChemSusChem 2014, 7, 3442–3451.
[40] J. Rouquerol, D. Avnir, C. W. Fairbridge, D. H. Everett, J. M. Haynes, N.
Pernicone, J. D. F. Ramsay, K. S. W. Sing, K. K. Unger, Pure Appl. Chem.
1994, 66, 1739–1758.
[41] G. Jander, K. F. Jahr, G. Schulze, J. Simon, Maßanalyse: Theorie und Praxis
der Titrationen mit chemischen und physikalischen Indikationen, De
Gruyter, Berlin, 2003.
[42] J. Luo, Q. Zhang, S. L. Suib, Inorg. Chem. 2000, 39, 741–747.
[26] P. Rasiyah, A. C. Tseung, J. Electrochem. Soc. 1984, 131, 803–808.
[27] I. Ursu, R. Alexandrescu, I. N. Mihailescu, I. Morjan, V. Jianu, C. Popescu,
J. Phys. D 1986, 19, 1183.
Received: April 8, 2015
Published online on September 1, 2015
Chem. Eur. J. 2015, 21, 14958 – 14968
www.chemeurj.org
14968
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim