Ionic-liquid-mediated active-site control of MoS2 for the electrocatalytic hydrogen evolution reaction

Article abstract of DOI:10.1002/chem.201200255

The layered crystal MoS₂ has been proposed as an alternative to noble metals as the electrocatalyst for the hydrogen evolution reaction (HER). However, the activity of this catalyst is limited by the number of available edge sites. It was previously shown that, by using an imidazolium ionic liquid as synthesis medium, nanometre-size crystal layers of MoS₂ can be prepared which exhibit a very high number of active edge sites as well as a de-layered morphology, both of which contribute to HER electrocatalytic activity. Herein, it is examined how to control these features synthetically by using a range of ionic liquids as synthesis media. Non-coordinating ILs with a planar heterocyclic cation produced MoS₂ with the de-layered morphology, which was subsequently shown to be highly advantageous for HER electrocatalytic activity. The results furthermore suggest that the crystallinity, and in turn the catalytic activity, of the MoS₂ layers can be improved by employing an IL with specific solvation properties. These results provide the basis for a synthetic strategy for increasing the HER electrocatalytic activity of MoS₂ by tuning its crystal properties, and thus improving its potential for use in hydrogen production technologies. Copyright

Full text of DOI:10.1002/chem.201200255

DOI: 10.1002/chem.201200255
Ionic-Liquid-Mediated Active-Site Control of MoS for the Electrocatalytic
2
Hydrogen Evolution Reaction
[
a]
[a]
[b]
Vincent Wing-hei Lau, Anthony F. Masters, Alan M. Bond, and
[
a]
Thomas Maschmeyer*
Abstract: The layered crystal MoS has
which contribute to HER electrocata-
lytic activity. Herein, it is examined
how to control these features syntheti-
cally by using a range of ionic liquids
as synthesis media. Non-coordinating
ILs with a planar heterocyclic cation
produced MoS2 with the de-layered
morphology, which was subsequently
shown to be highly advantageous for
HER electrocatalytic activity. The re-
sults furthermore suggest that the crys-
tallinity, and in turn the catalytic activi-
2
been proposed as an alternative to
noble metals as the electrocatalyst for
the hydrogen evolution reaction
(
HER). However, the activity of this
ty, of the MoS layers can be improved
2
catalyst is limited by the number of
available edge sites. It was previously
shown that, by using an imidazolium
ionic liquid as synthesis medium, nano-
metre-size crystal layers of MoS2 can
be prepared which exhibit a very high
number of active edge sites as well as
by employing an IL with specific solva-
tion properties. These results provide
the basis for a synthetic strategy for in-
creasing the HER electrocatalytic ac-
tivity of MoS2 by tuning its crystal
properties, and thus improving its po-
tential for use in hydrogen production
technologies.
Keywords: electrochemistry ·
heterogeneous catalysis · hydrogen ·
ionic liquids · molybdenum sulfide
a
de-layered morphology, both of
Introduction
economical process of hydrogen production, the electrocata-
lyst needs to be sufficiently active while either minimising
the use of noble metals, or employing non-precious alterna-
tives. Furthermore, intricate coordination or organometallic
catalysts should be avoided, as the increased complexity of
the overall catalyst system will lead to a higher cost of cata-
lyst preparation. Thus, hydrogen-evolving catalysts such as
hydrogenase and nitrogenase are unsuitable for large-
scale applications, despite the fact that they are highly active
and employ inexpensive metals.
Hydrogen will become an increasingly important commodity
in the future, since it has been proposed as the principal
energy carrier in the hydrogen-economy paradigm. Pres-
ently, the dominant method for hydrogen production is
alkane reforming, a process that usually depends on non-re-
[1]
[3]
[4]
newable fossil-based feedstocks and releases CO and CO as
2
environmentally harmful by-products. Furthermore, the
presence of CO is undesirable even in trace amounts, as it
can poison the noble metal electrocatalysts in a hydrogen
fuel cell. Hydrogen production by the electrochemical split-
ting of water, therefore, offers a cleaner alternative (provid-
ed that a renewable source of electricity is used), as this pro-
cess yields only hydrogen and oxygen as products, and be-
cause the availability of the precursor, water, is essentially
limitless. However, for this reaction to proceed at a satisfac-
tory rate, expensive electrocatalysts are often required. For
example, the best electrocatalysts for one of the half-reac-
tions of water splitting, the hydrogen evolution reaction
The HER electrocatalyst system investigated here is the
layered crystal MoS , the edges of which bear a close resem-
2
blance both structurally and electronically to the edges of
[5]
the active centre of the nitrogenase enzyme. The activity
of MoS for HER has indeed been experimentally verified:
2
ꢀ
2
ꢀ1
MoS has an activity of 1.64ꢀ10
active site. Since the HER activity of MoS is only 57
times lower than that of platinum (9.4ꢀ10
s
based on turnover per
2
[6]
2
ꢀ1
ꢀ
1
s
for the Pt-
AHCTUNGTR(ENNNUG 111) surface plane), whilst being more than 1000 times
cheaper (based on the price of the molybdenum and plati-
+
ꢀ
(
HER, 2H +2e !H ), are platinum and the other noble
num), MoS has been proposed as an inexpensive candidate
2
2
[2]
metals. Hence, for the water-splitting reaction to be an
catalyst for HER. However, the activity of MoS is limited
2
by the population of its active edges, which only constitute
a small proportion of the total surface area. The majority of
the molybdenum and sulfur atoms occupy the interior of the
crystal layer to form the basal planes, which account for the
majority of total surface area but are catalytically inactive.
[
a] Dr. V. W.-h. Lau, Prof. A. F. Masters, Prof. T. Maschmeyer
Laboratory of Advanced Catalysis for Sustainability
School of Chemistry,
University of Sydney, NSW, 2006 (Australia)
Fax : (+61)2-9351-3329
To explore the potential of MoS as a practical HER elec-
2
E-mail: th.maschmeyer@chem.usyd.edu.au
trocatalytic alternative to the noble metals, it is necessary to
improve its activity by maximising the number of active
edge sites.
[
b] Prof. A. M. Bond
School of Chemistry, Monash University
Clayton, Victoria, 3800 (Australia)
8230
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 8230 – 8239

FULL PAPER
One route to increasing the number of active edges is to
decrease the size of the MoS layers to nanometre dimen-
2
sions. We previously reported such a synthetic route where-
by MoS crystals were prepared by thermal decomposition
2
of (NH ) MoS in 1-alkyl-3-methylimidazolium triflate ionic
4
2
4
[7]
liquids (ILs) [C C IM] ACHUTNGRNEUNG[ OTf]. The choice of an IL as sol-
1
n
vent was based on the observation that active surfaces in
crystalline catalysts are usually under-coordinated high-
[8]
energy surfaces, and that ILs have been reported to stabi-
lise highly energetic, thermodynamically unstable surface
[9]
features in crystals. In this study, the IL-synthesised MoS2
samples not only displayed greater electrocatalytic activity
for HER than the control samples, but also exhibited several
important morphological changes. Specifically, the IL-syn-
thesised samples consisted of smaller and more unstacked
Figure 1. Structures and notations of the three aromatic and the three
non-aromatic ILs. The counterion common to all ILs is the trifluoro-
AHCTUNGERTNNUNGm ethanesulfonate anion.
[12a,e]
MoS layers compared to the controls. The origin of this un-
ly favourable.
To test whether unstacking/de-layering is
2
stacked/de-layered morphology demands further investiga-
tion, as it may impact on the catalytic activity in several
ways. Firstly, TEM analysis suggests that such morphology
caused by the aromaticity of the IL, three aromatic and
three non-aromatic ILs (Figure 1) were employed as solvent
for the synthesis of MoS . As several of these ILs lack physi-
2
increases the exposure of the MoS layer edges, and may im-
cochemical data in the literature, they were thoroughly char-
acterised by NMR, ESI-MS and, in particular, thermogravi-
metric analysis (TGA) to ensure that they have the required
thermal stability for the synthesis. The six ILs were used to
2
prove diffusion of substrate and product to and from the re-
action centres. Secondly, this de-layered morphology is
known to affect the relative rates of hydrodesulfurisation
and hydrogenation in aromatic sulfur-containing com-
prepare six samples of MoS by thermal decomposition of
2
[10]
pounds. This morphology–selectivity relationship is ration-
(NH ) MoS following a modified version of the protocol
4
2
4
[10a]
[7]
alised by the rim-edge model,
which proposed that hy-
described previously. The six samples are denoted by the
drogenation occurs exclusively on the rim sites, that is, the
edges on the top and bottom layers of the MoS2 stacks,
whereas sulfur hydrogenolysis is proposed to occur on both
the rim and edge sites, the latter of which are all of the edge
sites on the layers between the top and bottom of the stacks.
One hydrodesulfurisation/hydrogenation study has suggest-
ed that the inflection points at the curvature of unsupported
solvent employed in the synthesis. These MoS samples were
2
characterised morphologically and evaluated for their HER
electrocatalytic performance, with particular focus on how
the differing morphologies affected their catalytic activity.
Results and Discussion
MoS layers may also act as the hydrogenation active site,
2
and that the number of these points increases as the MoS
layers become more unstacked. Given the number of po-
tential effects this de-layered morphology has on the catalyt-
Characterisation: The procedure for the synthesis of MoS2
2
[11]
involved a more rigorous purification step than that report-
[7]
ed previously. Specifically, the black MoS powder samples
2
ic properties of MoS , it is of fundamental and practical in-
were washed with ethanol, water, acetone and hexane (as
opposed to only with ethanol) to ensure complete removal
of the reactants and the IL. The absence of strong aliphatic
CH stretching bands in the FTIR spectra (Figure 2) in the
2
terests to systematically elucidate this morphology–activity
relationship for HER electrocatalysis. Thus, the objectives
of the present study were to develop a method of syntheti-
cally controlling this de-layered morphology and investigate
the aforementioned morphology–activity relationship, with
the ultimate goal of exploiting any benefit to the catalyst ac-
ꢀ
1
region between 3000 and 2000 cm for all six samples is
consistent with removal of the IL from the MoS powder.
2
Other IR features are observed in the region between 2000
ꢀ
1
ꢀ1
tivity arising from unstacking of the MoS layers.
and 1000 cm , specifically a broad band at 1600 cm , corre-
2
[13]
One potential cause of de-layering in the IL-synthesised
sponding to adsorbed water on air-contacted MoS ,
bands centred around 1500 and 1300 cm , which correspond
to removable surface-bound ammonia from decomposition
and
2
ꢀ1
MoS is that the aromaticity of the imidazolium IL disrupts
2
the van der Waals interactions holding the MoS layers to-
2
[14]
gether. Several literature examples support this “de-layer-
ing-via-aromaticity” hypothesis and it is one that will be ex-
of (NH ) MoS .
Moreover, as the MoS2 samples were
4
2
4
stored in air, bands attributable to surface oxide species
[12]
ꢀ1 [13]
amined closely in this study.
For example, imidazolium
were evident in the region of 1100–1000 cm . This molyb-
ILs have been reported to untangle bundles of carbon nano-
denum oxide species on the surface is the predominant fea-
ture in all six samples, as observed in their Raman spectra
at 514 nm excitation (Figure 3), where the phonons corre-
[12b,c]
[12d]
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
tubes
and exfoliate graphite into graphene;
these
phenomena were attributed to disruption of the graphitic p–
p interactions by the aromatic cation. Moreover, adsorption
of conjugated and aromatic compounds onto the basal
[15]
sponding to MoO
are more intense than those of
3
[16]
MoS2. This observation indicates that the surface is princi-
pally composed of the oxide, likely due to the large amount
planes of the MoS layers has been shown to be energetical-
2
Chem. Eur. J. 2012, 18, 8230 – 8239
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8231

T. Maschmeyer et al.
Figure 2. FTIR spectra of the MoS
2
samples; the sharp signals at
ꢀ
1
Figure 4. XRD patterns of the six MoS
2
samples, indexed to single-layer
ꢁ
2400 cm correspond to atmospheric carbon dioxide, while the broad
signal centred at ꢁ3200 cm corresponds to moisture from the atmos-
[
17]
ꢀ1
2H-MoS
2
.
2
The pattern of a commercial sample of MoS from Aldrich
is shown for comparison.
phere or from the KBr matrix. The spectrum of a commercial sample of
2
MoS from Aldrich is shown for comparison.
cant height, and that stacks do not appear frequently in the
sample. The nanometre-scale arrangement of the MoS2
layers and bulk morphology of these six samples were ana-
lysed by TEM and SEM (Figures 5 and 6, respectively).
With the exception of MoS -[C Pyz] ACHUTNGRNENUG[ OTf], SEM analysis
2
4
shows that all of the MoS samples have a spherical mor-
2
phology with diameters of 200–300 nm, indicative of a radial
[19]
growth mechanism from well-solubilised precursors.
The
surfaces of these spherical particles consist of MoS layers of
2
differing degree of unstacking/de-layering, as observable
under TEM analysis. This de-layered morphology is most
pronounced in MoS -[C Py] ACTUHTGNRENNUGN[ OTf] and MoS -[C C IM] ACHTUNGTRENNUNG[ OTf],
2 1 4
2
4
in which the MoS layers are peeling away from the surface
2
of the spherical particles. This morphology, however, is also
observable in MoS -[C C Pyr] ACHUTNRGNENUG[ OTf], in contradiction to the
2
1
4
de-layering-via-aromaticity hypothesis. The de-layered mor-
phology is present to a much lesser extent in MoS2-
Figure 3. Raman spectra of the MoS
MoS phonons). The spectrum of a commercial
phonons; ^: MoO
sample of MoS from Aldrich is shown for comparison.
2
samples at 514 nm excitation (!:
[
C C MO] ACHTUNGTREGNNNU[ OTf] and MoS -[C C PI] ACHTUERNTGNUGN[ OTf]. MoS -[C Pyz]-
1 4 2 1 4 2 4
2
3
AHCTUNGTRENNUGN[ OTf] appears to be an outlying result: SEM showed that
2
the sample has both nanometre-sized spheres, indicative of
radial crystallisation from solubilised precursor, and micro-
of reactive surfaces susceptible to atmospheric oxidation.
The bulk of the samples, however, is definitively composed
of MoS rather than MoO , given the absence of the MoO
3
reflections in their XRD patterns, as shown in Figure 4.
The two key features in the XRD patterns are the reflec-
meter-sized prisms similar to bulk MoS . This bimodal mor-
2
phology can be attributed to the presence of both solubilised
and non-solubilised precursors, which leads to a bifurcation
in morphological outcome. It is uncertain why this IL is less
efficient in solubilising the thiomolybdate precursor than
the other five ILs. However, this is the only IL in the set
that has an electron lone pair, which can affect the intermo-
lecular properties of this solvent. Furthermore, TEM analy-
sis showed that neither of these two morphologies exhibited
2
3
tions corresponding to single-layer MoS sheets (i.e., intra-
2
[17]
layer ordering)
and the absence of the (002) reflection,
the latter corresponding to the 6 ꢂ inter-layer spacing of
stacked MoS . Amongst the IL-synthesised MoS samples,
2
2
MoS -[C C IM]
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
[OTf] and MoS -[C Py]
A
H
N
T
E
N
N
[OTf] have higher
unstacking of the MoS layers. One explanation is that the
2
1
4
2
4
2
crystallinity than the others, as evident from their sharper
100) and (110) intra-layer reflections. The higher crystallini-
IL–MoS interaction(s) required to unstack the layers were
2
(
not present, but instead other interactions, especially coordi-
nation bonding, were more favourable and hence dominant.
This explanation is supported by the fact that aromatic het-
erocycles containing multiple nitrogen atoms (e.g., pyrimi-
dine and pyrazine) are known to form stable complexes
with molybdenum over a number of oxidation states (4–
ty of these samples is an important factor when evaluating
their electrocatalytic activity for HER, since crystallinity af-
[18]
fects the formation of well-defined active edge sites. The
absence of the (002) reflection in all six IL-synthesised sam-
ples suggests that the MoS layers do not stack to a signifi-
2
8232
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 8230 – 8239

Active Site Control of MoS2
FULL PAPER
Figure 6. SEM at 6000 magnification of MoS
2
synthesised in ILs.
[OTf]. d) [C Py][OTf].
ACHTUNGTRENUN[NG OTf]. The scale bar in each micrograph cor-
a) [C
e) [C
1
C
C
4
PI]
IM]
A
H
U
G
R
N
U
G
1
C
4
MO]
A
H
U
T
E
N
N
[OTf]. c) [C
1
C
4
Pyr]
A
H
U
G
R
N
U
G
4
ACHTUNGTRENNUNG
1
4
A
T
G
R
N
U
G
4
Pyz]
responds to 5 mm.
To meaningfully rank the electrocatalytic performance,
the BET surface areas were characterised as well by nitro-
gen sorption (Table 1). It is evident that the de-layered mor-
phologies correlate with the BET surface area: the three
samples with the de-layered morphology—MoS -[C Py]-
5
Figure 5. TEM at 10 magnification of MoS
2
synthesised in ILs.
[OTf]. d) [C Py][OTf].
a) [C
e) [C
1
C
C
4
PI]
IM]
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
[OTf]. b) [C
1
C
4
MO]
ACHTUNGTRENNUNG
A
H
U
G
R
N
N
[OTf]. c) [C
1
C
4
Pyr]
A
H
U
T
E
N
G
4
ACHTUNGTRENNUNG
2
4
1
4
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
[OTf]. f) [C
4
Pyz][OTf].
A
H
U
G
R
N
U
G
A
H
U
G
R
N
U
G
A
H
U
G
R
N
U
[OTf]—all
2
1
4
2
1
2
4
ꢀ1
have surface areas in the hundreds of m g , whereas the
samples without this morphology have values below
[
20]
6
), that is, this IL may co-ordinate to both the thiomolyb-
date precursor and the MoS product.
30 m g . The results are consistent with the fact that, when
2
For the remaining five MoS samples, the evidence here
the MoS layers are in an unstacked arrangement, there is
2
2
demonstrates that not only is the de-layered morphology ob-
greater exposure of the top and bottom basal surfaces of the
served when MoS is synthesised in non-coordinating aro-
MoS layers that are accessible by the sorbent molecules.
2
2
matic ILs in agreement with the hypothesis, but also that
this morphology is observed when the aliphatic IL
[
C C Pyr] ACHTUNGTRENNUNG[ OTf] is employed. Presumably, heterocycles of
1 4
Table 1. BET surface areas of the MoS
sorption branch of the nitrogen sorption isotherm at 77 K. The surface
2
samples, calculated from the ad-
fewer than five members are sufficiently planar to interact
with the MoS basal surfaces and prevent the layers from
stacking together, while those with six or more members
area of a commercial sample of MoS
son.
2
from Aldrich is shown for compari-
2
2
ꢀ1
(
morpholinium and piperidinium) are less efficient for this
BET surface area [m g
]
role, due to ring buckling in the boat/chair conformation.
These results indicate that synthetic control over this un-
stacking/de-layering trend is achievable by employing ILs
with the desired intermolecular interactions as synthesis
media. This morphological control thus enables an investiga-
tion into the structure–functionality relationships between
the different morphologies and the HER electrocatalytic ac-
control
MoS -Aldrich
8.3
2
MoS
MoS
MoS
2
2
2
-[C
-[C
-[C
1
4
4
C
4
IM]
Py][OTf]
Pyz][OTf]
A
T
N
T
E
N
N
[OTf]
203
147
15.3
aromatic
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
MoS
MoS
MoS
2
2
2
-[C
-[C
-[C
1
1
1
C
4
C
4
C
4
MO]
PI][OTf]
Pyr][OTf]
A
H
U
G
R
N
U
G
18.7
29.1
201
non-aromatic
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
tivities of the various MoS samples.
2
Chem. Eur. J. 2012, 18, 8230 – 8239
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8233

T. Maschmeyer et al.
HER electrocatalysis: HER electrocatalytic activities of the
six MoS samples were evaluated with a rotating electrode
2
under acidic conditions by two methods: 1) the standard
Levich–Koutecky experimental protocols for the determina-
tion of the limiting current I ; and 2) time-based single-po-
K
tential coulometry for estimating the turnover frequency
(
TOF), assuming a 100% Faradaic efficiency for HER and
that all 10 mg of MoS powder on the electrode participated
2
in the electrocatalytic reduction. Polarisation curves and the
corresponding Tafel plots are shown in Figure 7, Levich–
Koutecky plots in Figure 8 and time–current plots in
Figure 8. a) Levich–Koutecky plot of the HER current at ꢀ533 mV and
ꢀ
1
b) the same plot reproduced in the inverse-current range of 0–6000 A
.
As turbulent flow or edge effects manifest at rotation rates above
600 rpm (below an inverse square root of rotation rate of 0.077), the
limiting current I (y intercept) was determined by linear fitting of the
1
K
data points at scan rates between 400 and 1600 rpm (inverse square root
of rotation rate 0.77–0.16).
of MoS -[C Pyz]
consistent with nanometre-sized MoS platelets. Tafel plots
A
H
U
G
R
N
N
[OTf], is approximately ꢀ200 mV, a value
2
4
[6]
2
of the IL-synthesised MoS2 samples, not including MoS2-
[
C Pyz]AHCNUTGTERNNUNG
4
Figure 7. a) Linear-scan voltammogram at 1000 rpm and b) the corre-
sponding Tafel plot.
2
Table 2. Electrochemical data of the six IL-synthesised MoS samples, compared to those of the two controls.
Figure 9. I , Tafel slopes and
K
I
K
at ꢀ533 mV
Tafel slope below
Tafel slope above
TOF
ACHTUNGTRENNUNG[ 10 s ]
TOFs for the six IL-synthesised
MoS2 samples, together with
those of the two controls—
3
ꢀ1
[
mA]
300 mV overpotential
300 mV overpotential
ꢀ
1
ꢀ1
[mVdecade
]
[mVdecade
]
GCE
Aldrich
30.8
34.4
–
–
111
112
3.9
2.2
control
MoS -[Aldrich] and the un-
2
modified glassy carbon elec-
trode (GCE)—are summarised
in Table 2.
C
C
C
1
4
4
C
Py
Pyz
4
IM
529
370
37.8
156
154
–
323
334
205
48
32
3.7
aromatic
From
the
voltammetric
curves, the onset potential of
HER for all of the IL-synthes-
C
C
C
1
C
1
C
1
C
4
4
4
Pyr
PI
MO
382
243
245
114
161
147
481
483
401
25
11
11
non-aromatic
ised MoS , with the exception
2
8234
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 8230 – 8239

Active Site Control of MoS2
FULL PAPER
Figure 9. Single-potential time-based plot with the working electrode held at ꢀ533 mV at 1000 rpm. Each plot was integrated from the region where the
current reached a plateau value to yield the number of charges passed. Turnover frequency was estimated by assuming a Faradaic efficiency of 100%
2
and that all 10 mg of the MoS powder deposited onto the electrode was involved in the electrocatalysed reaction.
[
22]
the five samples yielded Tafel slopes between 114–
such as MoO , are not only inactive for HER, but also
3
ꢀ
1
1
61 mVdecade . Literature values for the Tafel slope are
electrochemically unstable and tend to be reduced under
the acidic conditions and the potential applied in this work,
in accordance with the Pourbaix diagram for molybdenum
ꢀ
1
either 60 or 120 mVdecade , depending on the preparative
method of MoS2. Nonetheless, the values obtained in this
work are close to the theoretical slope of 120 mVdecade
for HER electrocatalysts, in which adsorption of protons is
the rate-limiting step. This mechanism is consistent with
[6]
ꢀ
1
[23]
oxide.
While it is not certain whether the surface after
electrochemical reduction is terminated by molybdenum (re-
moval of surface oxygen atoms) or sulfur (removal of mo-
lybdate-like species), the steady currents observed in the
single-potential time-based plot indicate that the resulting
surface is catalytically active for HER.
The order of activity for the MoS electrocatalysts showed
a clear correlation between the morphology and the HER
electrocatalytic performance of the MoS samples. The three
most active samples consisted of spherical particles with di-
ameters of 100–300 nm and exhibited the unstacked/de-lay-
[21]
the position of MoS in the volcano plot where the Gibbs
2
free energy of forming a surface catalyst–hydrogen bond is
[6b]
positive, that is, non-spontaneous hydrogen adsorption in
which the surface coverage of the catalyst with hydrogen
does not reach saturation. In the second region above an
overpotential of 300 mV, the five samples yielded slopes in
2
2
ꢀ
1
excess of 300 mVdecade . This is attributed to diffusion
limitations of the substrate through the voids between the
nanometre-sized particles. The outlying sample, MoS2-
ered morphology. Both MoS -[C C PI] ACHTUNGTERNN[UNG OTf] and MoS -
2
2
1
4
[
C Pyz]
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
[OTf], displayed essentially similar electrocatalytic
[C C MO] ACHTUNRTEGNNUN[G OTf], which also consisted of spherical particles
1 4
4
properties to the two controls, MoS -[Aldrich] and the
with similar diameters but without the de-layered morpholo-
gy, exhibited lower activity. The least active electrocatalyst
2
GCE, with an onset potential of around ꢀ400 mV. Its Tafel
slope is twice that of the two controls, and may also be at-
tributed to diffusion limitations, given that this sample has
is MoS -[C Pyz] ACTHUNGTERNN[UNG OTf], which has identical activity to MoS -
2
2
4
[Aldrich]. This result is not unexpected since, from SEM
analysis, MoS -[C Pyz][OTf] consisted of predominantly
much finer particles than MoS -[Aldrich].
ACHTUNGTRENNUNG
2
2
4
The order of activity of the MoS samples, based on both
prism-shaped particles, which is the identical morphology to
2
IK from the Levich–Koutecky plot and the TOF from coul-
the commercial MoS sample. Therefore, this result clearly
2
ometry, is MoS -[C C IM]
A
H
U
G
R
N
U
G
A
H
U
G
R
N
U
G
demonstrates that the de-layered morphology is beneficial
for HER electrocatalysis, due either to better exposure of
the active edges and therefore improved substrate/product
transfer to and from the active sites, or to the presence of
additional active sites in the form of the inflection points at
the curvature of the MoS2 layers. The quantification of
active sites in future experiments may elucidate which of
the two is the major contributor to the increase in activity.
Furthermore, there are significant differences in electroca-
2
1
4
2
4
2
[
C C Pyr]
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
[OTf]>MoS -[C PI]
1
4
2
1
4
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
[OTf]>MoS -[C Pyz]
A
H
U
G
R
N
N
[OTf]ꢁMoS -[Aldrich]ꢁGCE. From
2
4
the time–current plot, the electrocatalytic activity is sus-
tained over a one-hour period for all MoS2 samples; for
some of the MoS samples, this indicates that the electro-
2
ACHTUNGTRENNUNGc atalysts are active and stable for more than one hundred
turnovers. The presence of the surface oxide, as observed in
the Raman spectra above, is not expected to affect the elec-
trocatalytic HER activity, because oxides of molybdenum,
talytic activity amongst the three MoS samples with de-lay-
2
Chem. Eur. J. 2012, 18, 8230 – 8239
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8235

T. Maschmeyer et al.
ered morphology. If one assumes that the extent of Table 3. Solvent properties under ambient conditions of the triflimide ILs that yielded
the de-layered morphology in MoS
2
.
de-layering is correlated with the BET surface area
due to exposure of the basal planes), the results in-
Dielectric
constant e
Reichardtꢃs E
polarity scale
T
[b]
Kamlet–Taft dipolarity/
polarisability parameter p*
(
[
a]
[c]
dicate that there is an additional factor that contrib-
uted to either the higher activity of MoS -[C C IM]-
A
H
U
G
R
N
U
G
1
4
1
C
4
IM]
Py][NTf
2
Pyr][NTf ]
A
H
U
G
R
N
U
G
2
]
11.6
11.5
11.9
0.64
0.65
0.55
0.955
0.972
0.971
2
1
4
A
H
U
G
R
N
U
A
H
U
G
R
N
U
G
ACHTUNGTRENNUNG[ OTf] over MoS -[C C Pyr] AHCTUNGTRUNNENG
2 1 4
A
H
N
T
N
N
C
4
activity of MoS -[C Py]ACHTUGNTRENNNUG
2
4
[
26]
[27]
[
a] Literature values from Krossing et al.
[b] Literature values from Reichardt.
2
1
4
[28]
[
c] Literature values from DꢃAnna et al.
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
[OTf], despite having a significantly lower BET sur-
face area. One distinguishing feature of MoS2-
C C IM][OTf] and MoS -[C Py][OTf] is that their intra-
[
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
A
H
U
G
R
N
N
with the catalyst activity of the de-layered MoS sample, but
1
4
2
4
2
layer crystal reflections in their XRD patterns are more de-
not with the crystallinity of the product. Presently it is un-
certain whether this correlation has a causal relationship. It
is, however, certain that the mechanism proposed by Wu
et al., regarding the supersaturation due to reduced solubili-
ty of the precursor in solvent with low dielectric constant, is
not a satisfactory explanation for the present case. Since the
fined than those arising from the other four MoS samples,
2
that is, these two samples are more crystalline. Since crystal-
linity is desirable for the formation of well-defined active
[18]
edge sites, it is the factor to which the improved catalytic
performance of these two samples is attributed. Given that
the properties of the synthesis media are known to affect
the crystal properties, it is noteworthy that both non-coordi-
[30]
thiomolybdate anion is a hydrogen-bond acceptor, hydro-
gen bonding is considered to be the dominant solvating in-
teraction for this precursor. Moreover, electrostatic models
of solvation (i.e., those based on physical constants such as
dielectric constant, permanent dipole moment and refractive
index) often fail in rationalising solvent effects, as they do
nating aromatic ILs yielded MoS with better crystallinity
2
than the non-aromatic ILs of similar molecular size. This
effect may be rationalised by the examination of several lit-
erature postulates. For example, Wu et al. claimed that the
crystallinity (in terms of crystal size and lattice strain) of
[31]
not account for specific solvent/solute interactions.
One
TiO , and hence its photocatalytic activity, can be controlled
postulate for the improved crystallinity is that ILs with
higher hydrogen-bond acidity are better able to solubilise
the precursor or crystal nuclei. This may lead to a controlled
crystallisation or growth process, and thus afford products
with either a reduced number of crystal defects or well-or-
dered edge sites. Obviously, a number of other solvent prop-
erties may also affect the outcome of the crystallisation, and
these include solvent viscosity, steric hindrance and surface
tension. It is expected that future investigations into the
2
by using solvents of different di ACHTUGNTRENNUeGN lectric constants in a solvo-
thermal synthetic route, with solvents of small dielectric
[24]
constant yielding products of higher crystallinity. They at-
tributed this solvent effect to a reduction in precursor solu-
bility as the dielectric constant of the solvent is reduced.
This, they claimed, leads to a supersaturation of crystal
nuclei and faster crystal growth and, through Ostwald ripen-
ing, yields product of larger crystal size and lower lattice
strain. We also recently showed that the activity of the pho-
tocatalyst CdS is heavily dependent on the solvating ability
of the synthesis medium, quantified by using the Reichardt
mechanism of MoS crystallisation and the mode of interac-
2
tions of the ILs with both the thiomolybdate precursor and
the MoS crystals will yield insights into methods of control-
2
[25]
solvatochromic betaine and ET polarity scale.
solvation properties of the ILs may explain the improved
crystallinity of MoS -[C C IM][OTf] and MoS -[C Py][OTf].
methanesulfonate ILs are
Thus, the
ling the degree of crystallisation.
A
H
U
G
R
N
N
ACHTUNGTRENNUNG
2
1
4
2
4
As solvent parameters of trifluoro
A
H
U
G
R
N
U
G
Conclusion
scarce in the literature, the solvation parameters of the cor-
responding N,N-bis(trifluoromethanesulfonyl)imide (NTf2)
ILs were instead examined. This comparison is justifiable,
The ability to predictably and synthetically manipulate crys-
tallinity and morphology of the layered crystal MoS2 is
highly desirable for catalytic applications, since the former
determines the formation of well-defined active sites, and
the latter affects the exposure and accessibility of substrates
to these sites. This article demonstrates that control over
these two properties can be exercised by judicious selection
of IL as solvent for thermal decomposition of (NH ) MoS
4
since both the OTf and NTf anions are weakly coordinating
2
and, where data are available, substitution of these two
anions does not significantly alter the solvating parameters
of the IL. The dielectric constant, the E value for polarity
T
and the Kamlet–Taft parameter p* for polarisability for the
three ILs are summarised in Table 3.
4
2
Consistent with the findings of Wu et al., both the
to MoS . It was shown that non-coordinating aromatic ILs
2
[
C C IM] and [C Py] ILs have lower dielectric constants and
yielded the most active MoS HER electrocatalysts, which
1
4
4
2
yielded products of better crystallinity. However, these two
ILs also exhibited better solvating ability compared to the
have high crystallinity and a de-layered/unstacked morphol-
ogy. Based on the present results, it was postulated that in-
termolecular interactions exerted by the IL govern the char-
[
C C Pyr] IL based on the E polarity scale, which measures
1
4
T
a combination of both hydrogen bonding acidity and polaris-
ability (p*). Another noteworthy observation is that the
polarisability parameter of the IL p* appears to correlate
acteristics of the MoS crystals, specifically hydrogen-bond
acidity for improving crystallinity, and aromatic or planar IL
molecular cations for the formation of the de-layered mor-
2
[29]
8236
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 8230 – 8239

Active Site Control of MoS2
FULL PAPER
1
phology. These findings provide the basis for several syn-
318.23 Hz,
CHCHN), 146.09 ppm (NCHCHCH);
77.55 ppm (CF S); MS (ESI +ve ion): m/z (%): 136 (100) [C
F
3
CS), 128.81 (CHCHCH), 144.69 (t,
J
CN =8.01 Hz,
1
9
F NMR (in CD
3
OD): d=
Py] , 421
thetic strategies to improve the activity of MoS HER elec-
2
+
ꢀ
3
4
trocatalysts, and may also have implications for preparing
+
ꢀ
(
45) [C
4
Py]
2
A
H
U
G
R
N
U
G
more active MoS catalysts for the industrially important hy-
2ꢀ
3ꢀ
2
(33) [C Pyr] ACHUTNTGRENNUN[G OTf] , 719 (30) [C Pyr]2 ACHUTGNERNNUG
4
4
drodeoxygenation and hydro ACHUTNRGNENUGd esulfurisation reactions.
(
ACHTUNGTRENNUNG
Experimental Section
AHCTUNGTRENNUNG
1
2
zole. Yield: 76%. Tdecomp =3198C; H NMR (in D O): d=0.98 (t, 3H,
All chemicals were of reagent grade and used without further purifica-
tion. Water was purified with an Elix Millipore 100 water purification
system and deoxygenated by purging with nitrogen gas overnight. The
3
3
J
HH =7.39 Hz, CH
3
CH
2
), 1.42 (sxt, 2H,
CH CH
), 8.54 (m, 1H, CHCHNNCH
HH =3.95 Hz, NCHCH), 9.74 ppm (d, 1H,
); C{ H} NMR (in CDCl
J
HH =7.92 Hz, CH
3
CH
2
CH
2
),
3
3
2
.12 (quin, 2H, JHH =7.72 Hz, CH
2
2
2
), 4.90 (t, 2H, JHH =7.52 Hz,
), 8.62 (m, 1H, CHCHNCH
NCH
9
2
CH
2
2
2
),
MoS
procedure.
2
precursor (NH
4
)
2
MoS
4
was synthesised according to a literature
3
3
.53 (d, 1H,
J
J
HH =5.77 Hz,
CH ), 19.30
), 120.56 (q, CF
CF =320.47 Hz), 136.56 (NNCHCH), 136.73 (NCHCH), 150.25
CH NCH), 154.13 ppm (CH NNCH); MS (ESI +ve ion): m/z (%):
36.67 (100) [C
[
32]
1
13
H (300.13 MHz) and C NMR (74.5 MHz) spectra were
13
1
CHNCH
CH CH
2
3
): d=13.30 (CH
3
2
recorded on a Bruker AVANCE 300 spectrometer. Sample solutions
were prepared in deuterated solvents (Cambridge Isotopes Laborato-
ries); spectra were referenced internally to the residual solvent absorban-
(
3
2
CH
2
), 32.16 (CH
2
CH
2
CH
2
), 65.94 (NCH CH
2
2
3
,
1
J
(
1
AHCTUNGTRENNUNG
2
2
[
31]
ces. ESI-MS spectra were collected on a Finnigan LCQ mass spectrom-
eter at the Mass Spectrometry Unit in the School of Chemistry, Universi-
ty of Sydney. Elemental analyses were performed at the Microanalytical
Unit at the Research School of Chemistry, Australian National Universi-
ty. CHN quantifications were performed on a Carlo Erba 1106 CHN ana-
lyser; fluorine and sulfur analyses were performed on a Dionex Ion
Chromatography Analyser. Thermogravimetric analyses were carried out
on a TA Instruments 2950 TGA. Each sample was loaded onto a platinum
+
+
4
Pyz] , 422.47 (20) [C
[OTf] ; MS (ESI ꢀve ion): m/z (%): 148.87 (100) [OTf] , 434.47 (7)
Pyz][OTf] ; elemental analysis (%) found (calcd): C 37.61 (37.76), H
.88 (4.58), N 9.48 (9.79), F 19.64 (19.91), S 10.69 (11.20).
4
Pyz]
2
A
H
U
G
E
N
N
[OTf] , 708.40 (13) [C
4
Pyz]
3
-
2+
ꢀ
2ꢀ
[
4
C
4
ACHTUNGTRENNUNG
Synthesis of N-butyl-N-methylpyrrolidinium trifluoromethanesulfonate:
An identical procedure to synthesising 1-butyl-3-methylimidazolium tri-
fluoromethanesulfonate was employed, using N-methylpyrrolidine in
1
ꢀ
1
ꢀ1
place of 1-methylimidazole. Yield: 87%. Tdecomp =3568C; H NMR (in
pan and decomposed under a N
2
stream (10 mLmin ) at 18C min
3
3
D
7
2
O): d=0.95 (t, 3H,
.42 Hz, CH CH CH ), 1.71 (m, 2H, CH
CH CH N), 3.04 (s, 3H, CH
H, Pyr, CH
9.77 (CH CH
CN =4.06 Hz, NCH
CN =2.95 Hz, pyr, NCH
J
HH =7.37 Hz, CH
3
CH
2
), 1.38 (sxt, 2H,
J
HH
=
from ambient temperature to 5008C.
3
2
2
2
CH
2
CH
2
), 2.21 (m, 4H, Pyr,
CH N), 3.53 ppm (m,
CN): d=13.20 (CH CH ),
), 25.63 (CH CH CH ), 48.52
CH ), 64.61 (t,
);
Synthesis of 1-butyl-3-methylimidazolium trifluoromethanesulfonate: The
title compound was prepared by anion exchange of 1-butyl-3-methylimi-
dazolium bromide with sodium trifluoromethanesulfonate. The bromide
2
2
3
N), 3.34 (m, 2H, CH
2
2
13
1
4
1
2
CH
2
N); C{ H} NMR (in CD
), 21.71 (Pyr, NCH CH
3
3
2
3
2
CH
2
2
2
2
2
2
salt was prepared by heating
3.9 mmol) and 1-methylimidazole (5.00 g, 60.9 mmol) under N
overnight. Unconverted substrate was removed by washing with hexane
4ꢀ50 mL), and the product was dried at 1008C in vacuo overnight. For
a
mixture of 1-bromobutane (8.76 g,
1
1
(
t,
J
J
3
), 64.32 (t,
CH
J
CN =2.79 Hz, NCH
2
2
6
2
at 408C
1
1
2
2
), 121.48 ppm (q,
J
CF =320.89 Hz, CF
3
19
F NMR (in CD
%): 142.2 (100) [C
3
OD): d=ꢀ77.64 ppm (CF
3
S); MS (ESI +ve ion): m/z
(
+
+
(
1
C
4
Pyr] , 432.87 (19) [C
1
C
4
Pyr]
2
A
H
U
G
R
N
U
anion exchange, the bromide salt (5.00 g, 22.8 mmol) in acetone (50 mL)
was mixed with sodium trifluoromethanesulfonate (4.12 g, 24.0 mmol) in
acetone (50 mL), and NaBr immediately formed as a white precipitate.
After stirring overnight, the mixture was filtered through Celite to
remove the NaBr precipitate and the acetone solvent was removed at
08C in vacuo. The resulting oil was re-dissolved in dichloromethane
200 mL) and washed with water in 5 mL aliquots until the aqueous
phase was free of bromide by testing for the formation of precipitate
with AgNO solution (0.3m). Typically, this step involved 4–6 washings to
ꢀ
2ꢀ
ion): m/z (%): 149.33 (100) [OTf] , 440.00 (66) [C
1
C
4
ACHTUNGTRENNUNG
mental analysis (%) found (calcd): C 40.85 (41.23), H 7.26 (6.92), N 4.79
4.81), F 16.04 (19.56), S 9.80 (11.00).
(
Synthesis of N-butyl-N-methylpiperidinium trifluoromethanesulfonate:
An identical procedure to synthesising 1-butyl-3-methylimidazolium tri-
fluoromethanesulfonate was employed, using N-methylpiperidine in
4
(
1
place of 1-methylimidazole. Yield: 67%. Tdecomp =3508C; H NMR (in
3
3
CDCl
.19 Hz, CH
NCH CH ), 3.10 (s, 3H, NCH
NMR (in CDCl ): d=13.36 (CH
3
): d=0.99 (t, 3H,
CH CH ), 1.82–1.96 (m, 2H, PI, NCH
), 3.32–3.48 ppm (m, 6H, CH
CH ), 19.44 (CH CH CH ), 19.86 (PI,
J
HH =6.75 Hz, CH
3
CH
2
), 1.42 (sxt, 2H,
CH CH ; m, 6H,
N); C{ H}
HH
J =
3
8
3
2
2
2
2
2
be halide-free. The dichloromethane phase was then washed with a fur-
ther 2ꢀ5 mL water to ensure minimisation of bromide impurity. The di-
chloromethane solvent was removed by heating at 1008C in vacuo over-
1
3
1
2
2
3
2
3
3
2
3
2
2
1
3
NCH CH ), 20.52 (PI, NCH CH CH ), 23.53 (NCH CH ), 47.35 (NCH ),
60.88 (PI, NCH ), 63.65 (NCH ), 120.65 ppm (q, CF , JCF =319.87 Hz);
MS (ESI +ve ion): m/z (%): 156 (100) [C C PI] , 461 (31, [C C PI] -
1 4 1 4 2
+
night. Yield: 80%. H NMR (in CD
3
CN): d=0.90 (t, 3H, JHH =7.36 Hz,
CH CH ), 1.82 (quin, 2H,
), 3.87 (s, 3H, NCH ), 4.17 (t, 2H,
N), 7.46 (m, H, CHCHN), 7.51 (m, H, CHCHN), 8.74 ppm
2
2
2
2
2
2
2
1
3
3
CH
J
3
CH
HH =7.50 Hz, CH
.28 Hz CH
s, H, NCHN); C{ H} NMR (in CD
CH CH CH ), 32.67 (CH CH CH ), 36.82 (NCH
22.08 (q, CF CF =320.60 Hz), 123.41 (NCHCH), 124.71 (NCHCH),
37.46 ppm (NCHN); F NMR (in CD
2
), 1.31 (sxt, 2H,
J
HH =7.47 Hz, CH
3
2
2
2
2
3
+
3
3
2
CH
2
CH
2
3
J
HH
=
ꢀ
AHCTUNGTRENNUNG
7
2
2
ꢀ
1
3
1
[
C
1
C
4
ACHTUNGTRENNUNG
(
(
3
CN): d=13.76 (CH
3
CH
2
), 20.02
CH ),
), 50.29 (NCH
2
2
H 7.52 (7.26), N 4.77 (4.59), F 18.76 (18.67), S 10.47 (10.50).
3
2
2
2
2
2
3
1
1
1
(
3
,
J
Synthesis of N-butyl-N-methylmorpholinium trifluoromethanesulfonate:
An identical procedure to synthesising 1-butyl-3-methylimidazolium tri-
fluoromethanesulfonate was employed, using N-methylmorpholine in
place of 1-methylimidazole. The purification step of the anion exchange
was modified, since the protocol involving the washing of the dichloro-
methane phase with water resulted in unacceptable yield. Instead, residu-
al bromide was removed by passing the dichloromethane solution
through an activated neutral alumina column. The eluent was found to
1
9
3
OD): d=ꢀ77.75 ppm (CF
3
S); MS
2
IM] -
+
ESI +ve ion): m/z (%): 138.73 (100) [C
1
C
4
IM] , 426.53 (37) [C
1
C
4
+
ꢀ
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
[OTf] ; MS (ESI ꢀve ion): m/z (%): 148.93 (100) [OTf] , 436.6 (16)
2
ꢀ
3ꢀ
1 4 1 4 2
[C C IM] ACHTUNGTRENNUNG[ OTf] , 724.60 (50) [C C IM] ACHTGNRUTNEUN[G OTf] .
Synthesis of N-butylpyridinium trifluoromethanesulfonate: An identical
procedure to synthesising 1-butyl-3-methylimidazolium trifluoromethane-
sulfonate was employed, using pyridine in place of 1-methylimidazole.
1
Yield: 81%. M.p. 33–368C; Tdecomp =3658C; H NMR (in D
2
O): d=0.96
HH =7.50 Hz,
), 4.63 (t, 2H,
N), 8.08 (t, 1H, JHH =6.82 Hz, CHCHCHN), 8.56
be bromide-free by the absence of precipitate when mixed with an aque-
3
3
1
(
t, 3H,
J
HH =7.34 Hz, CH
), 2.02 (quin, 2H, JHH =7.51 Hz, CH CH CH
3
3
CH
2
), 1.38 (sxt, 2H,
J
ous solution of AgNO
[D
3
(0.3m). Yield: 49%. Tdecomp =3498C; H NMR (in
3
CH CH
3
2
CH
2
2
2
2
6
]acetone): d=1.00 (t, 3H,
HH =7.32 Hz, CH CH CH
NCH ), 3.67 (m, 4H, NCH
OCH
(CH CH
J
HH =7.37 Hz, CH
CH
, Bu), 4.09 ppm (m, 4H,
]acetone): d=13.88 (CH CH ), 20.18
CH ), 47.01 (NCH ), 47.28 (NCH CH ),
3
CH
2
), 1.46 (sxt, 2H,
3
3
3
J
HH =7.29 Hz, CH
2
CH
2
J
3
2
2
), 1.90 (m, 2H, CH
, MO; m, 2H, NCH
); C{ H} NMR (in [D
CH ), 23.48 (NCH
2
2
CH
2
), 3.39 (m, 3H,
3
3
(
t, 2H,
J
HH =7.89 Hz, CHCHCH), 8.86 ppm (d, 2H,
J
HH =5.88 Hz,
CH ), 19.22
), 120.35 (q,
3
2
2
1
3
1
13
1
CHCHN);
CH CH CH
C{ H} NMR (in
D
2
O): d=13.17 (CH
3
2
2
6
3
2
1
(
3
2
2
), 33.10 (CH CH CH
2
2
2
), 62.27 (NCH
2
J
FC
=
3
2
2
2
2
3
2
2
Chem. Eur. J. 2012, 18, 8230 – 8239
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8237

T. Maschmeyer et al.
1
6
3
0.54 (MO, NCH
2
), 61.21 (MO, OCH
2
), 122.05 ppm (q, CF
3
,
J
CF
=
over frequency were estimated by integrating the time–current plot in
the region where the current is steady, assuming a Faradaic efficiency of
100% and that all 10 mg participated in the electrocatalytic reaction. All
experiments were repeated to verify the reproducibility of the catalyst
drop-casting process and the electrocatalytic activities.
+
21.87 Hz); MS (ESI +ve ion): m/z (%): 158 (100) [C
1
C
4
MO] , 465 (33)
+
ꢀ
[
C
1
C
4
MO]
36) [C
36.92), H 6.70 (6.82), N 4.34 (4.31), F 17.45 (17.52), S 9.85 (9.85).
: The ionic liquid (3.00 g) was
dried overnight at 1008C in vacuo before use. (NH MoS (150 mg) was
dissolved in the ionic liquid at 408C to form a deep red solution. This so-
2
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
[OTf] ; MS (ESI ꢀve ion): m/z (%): 149 (100) [OTf] , 456
2
ꢀ
(
(
1 4
C MO] ACHTUNGTRENNUNG[ OTf] ; elemental analysis (%) found (calcd): C 37.17
Synthesis and characterisation of MoS
2
4
)
2
4
ꢀ1
lution was then heated at 3008C under a H
for 2 h, yielding a black suspension. Ethanol was added to this suspension
and the black MoS powder and the ionic liquid were separated by cen-
trifugation. The MoS was further washed with ethanol (5ꢀ50 mL), fol-
lowed by water (2ꢀ15 mL), acetone (2ꢀ15 mL) and finally hexane (2ꢀ
2 2
/N flow (10%, 10 mLmin )
Acknowledgements
2
2
We thank the Australian Centre for Microscopy & Microanalysis for use
of their electron microscopes. Financial support from the University of
Sydney and the Australian Research Council is gratefully acknowledged.
1
8
5 mL). The solid was then dried in vacuo at ambient temperature. Up to
0% of the ionic liquid can be recovered by removing the ethanol in
vacuo.
Transmission electron microscopy was performed on a Philips CM120 Bi-
ofilter microscope operating at 120 kV. Images were digitally recorded on
a Gatan slow-scan CCD camera. Each sample was prepared by dispersing
the sample powder in ethanol and drop-casting onto a 200 mesh copper
grid coated with holey carbon (ProSciTech). XRD patterns were record-
ed with a PANalytical X’Pert PRO MPD X-ray diffractometer set up in
Bragg PDS mode with a PIXcel detector and CuKa (l=1.5419 ꢂ) as irra-
diation source. The samples were loaded on zero-background silicon
wafers on a spinning sample holder. Scans were carried out from 10 to
[1] D. A. J. Rand, R. M. Dell, Hydrogen Energy: Challenges and Pros-
pects, RSC Publishing, Cambridge, 2007.
[2] B. E. Conway, G. Jerkiewicz, Electrochim. Acta 2000, 45, 4075–4083.
[3] D. J. Evans, C. J. Pickett, Chem. Soc. Rev. 2003, 32, 268–275.
[4] S. C. Lee, R. H. Holm, Proc. Natl. Acad. Sci. USA 2003, 100, 3595–
3600.
[5] B. Hinnemann, P. G. Moses, J. Bonde, K. P. Jørgensen, J. H. Nielsen,
S. Horch, I. Chorkendorff, J. K. Nørskov, J. Am. Chem. Soc. 2005,
127, 5308–5309.
[6] a) J. Bonde, P. G. Moses, T. F. Jaramillo, J. K. Nørskov, I. Chorken-
dorff, Faraday Discuss. 2009, 140, 219–231; b) T. F. Jaramillo, K. P.
Jørgensen, J. Bonde, J. H. Nielsen, S. Horch, I. Chorkendorff, Sci-
ence 2007, 317, 100–102.
ꢀ
1
8
08 in 2q with step size of 0.0138 in 2q at a scan rate of 0.05258s . FTIR
spectra of the MoS samples were collected as NaBr matrices on a Varian
Scimitar 800 FTIR spectrometer with NaBr as spectroscopic background.
Raman spectra of the MoS samples were collected on a Renishaw
2
2
[
7] V. W.-h. Lau, A. F. Masters, A. M. Bond, T. Maschmeyer, Chem-
CatChem 2011, 3, 1739–1742.
Raman Invia Reflex spectrometer. The samples were excited with
a 514 nm argon laser at 10ꢀ objective and 10 s exposure for one scan in
ꢀ
1
[8] B. H. Davis, Handbook of Heterogeneous Catalysis, Wiley-VCH,
Weinheim, 2008.
the range of 100–3000 cm . Surface areas of the samples were measured
by nitrogen sorption analysis on a Micromeritics Accelerated Surface
Area & Porosimetry System (ASAP) 2020 instrument. The powder
sample was weighed out (50–70 mg) and outgassed at 908C at 7 mbar for
at least 4 h. Nitrogen adsorption and desorption isotherms were collected
at 77 K, and the surface areas were calculated by using the BET model
in the framework of the Micromeritics software. The BET surface area of
[
9] a) Z. Jiang, Q. Kuang, Z. Xie, L. Zheng, Adv. Funct. Mater. 2010, 20,
634–3645; b) Z. Ma, J. Yu, S. Dai, Adv. Mater. 2010, 22, 261–285;
3
c) T. Xu, X. Zhou, Z. Jiang, Q. Kuang, Z. Xie, L. Zheng, Cryst.
Growth Des. 2009, 9, 192–196; d) X. Zhou, Z. Xie, Z. Jiang, Q.
Kuang, S. Zhang, T. Xu, R. Huang, L. Zheng, Chem. Commun.
2
005, 5572–5574.
MoS
2
-[C
1
C
4
IM]
ly. This difference is attributed to the more rigorous purification steps
adopted in this work. As stated above, the MoS powder was washed
ACHTUNGTRENNUNG[ OTf] is significantly higher than that reported previous-
[
7]
[10] a) M. Daage, R. R. Chianelli, J. Catal. 1994, 149, 414–427; b) H.
Farag, K. Sakanishi, M. Kouzu, A. Matsumura, Y. Sugimoto, I.
Saito, J. Mol. Catal. A 2003, 206, 399–408; c) C. T. Tye, K. J. Smith,
Catal. Today 2006, 116, 461–468.
11] Y. Iwata, Y. Araki, K. Honna, Y. Miki, K. Sato, H. Shimada, Catal.
Today 2001, 65, 335–341.
12] a) U. Zimmermann, N. Karl, Surf. Sci. 1992, 268, 296–306; b) T. Fu-
kushima, A. Kosaka, Y. Ishimura, T. Yamamoto, T. Takigawa, N.
Ishii, T. Aida, Science 2003, 300, 2072–2074; c) T. Fukushima, T.
Aida, Chem. Eur. J. 2007, 13, 5048–5058; d) N. Liu, F. Luo, H. Wu,
Y. Liu, C. Zhang, J. Chen, Adv. Funct. Mater. 2008, 18, 1518–1525;
e) P. G. Moses, J. J. Mortensen, B. I. Lundqvist, J. K. Nørskov, J.
Chem. Phys. 2009, 130, 104709.
2
with, in addition to ethanol, water, acetone and hexane, which ensures
complete removal of trace IL and starting materials.
[
Electrochemical determination of HER activity: Hydrodynamic electro-
chemical experiments were performed on a BAS 100B Electrochemical
Workstation, connected to a BAS RDE-1 set-up. The working rotating
electrode was a 3 mm glassy carbon electrode (BAS), polished with alu-
mina (1 mm) on a Buehler pad according to BAS guidelines before use.
A platinum wire was used as the counter-electrode; the reference elec-
trode was an AgjAgCl electrode (3m NaCl, BAS). The reference elec-
trode was calibrated with ferri- and ferrocyanide; the potential for this
couple was taken as 436 mV versus the standard hydrogen electrode
[
[
[
[
13] F. Maugꢄ, J. Lamotte, N. S. Nesterenko, O. Manoilova, A. A. Tsyga-
nenko, Catal. Today 2001, 70, 271–284.
14] A. A. Tsyganenko, F. Can, A. Travert, F. Maugꢄ, Appl. Catal. A
(
SHE) at pH 7. All potentials henceforth are quoted against SHE. Modi-
fication of the working electrode was carried out as follows. The MoS
2
powder (10.0 mg) was weighed out on a Mettler Toledo XS105 balance,
and then dispersed with sonication in water (10.00 mL) to achieve a dis-
2
004, 268, 189–197.
15] D. Liu, W. W. Lei, J. Hao, D. D. Liu, B. B. Liu, X. Wang, X. H.
ꢀ
1
persion concentration of 1.00 mgmL . A 10 mL aliquot corresponding to
0 mg was drop-cast onto the glassy carbon electrode and dried under
Chen, Q. L. Cui, G. T. Zou, J. Liu, S. Jiang, J. Appl. Phys. 2009, 105,
1
0
23513.
a stream of nitrogen.
[
16] G. L. Frey, R. Tenne, M. L. Matthews, M. S. Dresselhaus, G. Dressel-
HER was performed in linear-scan voltammetry mode from 562 to
haus, Phys. Rev. B 1999, 60, 2883–2892.
[17] D. Yang, S. Jimꢄnez Sandoval, W. M. R. Divigalpitiya, J. C. Irwin,
R. F. Frindt, Phys. Rev. B 1991, 43, 12053–12056.
[18] J. H. Nielsen, L. Bech, K. Nielsen, Y. Tison, K. P. Jørgensen, J. L.
Bonde, S. Horch, T. F. Jaramillo, I. Chorkendorff, Surf. Sci. 2009,
603, 1182–1189.
[19] E. Matijevi c´ , Chem. Mater. 1993, 5, 412–426.
[20] a) D. Matoga, J. Szklarzewicz, J. Fawcett, Polyhedron 2005, 24,
1533–1539; b) S. Gupta, S. Pal, A. K. Barik, S. Roy, A. Hazra, T. N.
ꢀ
1
ꢀ1
ꢀ
637 mV at 5 mVs scan rate and 100 mA V sensitivity. The electrolyte
was aqueous H SO (60 mL, 0.500m). The rotation rate was varied from
00 to 6000 rpm and the current at ꢀ533 mV was used for the Levich–
Koutecky plot. The limiting current was determined by fitting the data
2
4
4
ꢀ
0.5
points to a linear correlation and extrapolating the current to w !0.
For evaluation of catalyst stability and turnover frequency, HER was per-
formed in single-potential, time-based mode. The electrode was rotated
at 1000 rpm and held at ꢀ533 mV for 1 h. Hydrogen production and turn-
8238
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 8230 – 8239

Active Site Control of MoS2
FULL PAPER
Mandal, R. J. Butcher, S. K. Kar, Polyhedron 2009, 28, 711–720;
c) S. Gupta, S. Roy, T. N. Mandal, K. Das, S. Ray, R. J. Butcher,
S. K. Kar, J. Chem. Sci. 2010, 122, 239–245; d) D. Matoga, J. Szklar-
zewicz, K. Lewi n´ ski, Polyhedron 2010, 29, 94–99.
[26] I. Krossing, J. M. Slattery, C. Daguenet, P. J. Dyson, A. Oleinikova,
H. Weingꢅrtner, J. Am. Chem. Soc. 2006, 128, 13427–13434.
[27] C. Reichardt, Green Chem. 2005, 7, 339–351.
[28] F. DꢃAnna, S. La Marca, P. Lo Meo, R. Noto, Chem. Eur. J. 2009, 15,
7896–7902.
[
21] a) S. Fletcher, J. Solid State Electrochem. 2009, 13, 537–549; b) A. J.
Appleby, J. H. Zagal, J. Solid State Electrochem. 2011, 15, 1811–
[29] Y. Marcus, Chem. Soc. Rev. 1993, 22, 409–416.
[
30] B. R. Srinivasan, S. N. Dhuri, M. Poisot, C. Nꢅther, W. Bensch, Z.
Naturforsch. B: Chem. Sci. 2004, 59, 1083–1092.
1
832.
[
[
[
[
22] Z. Chen, D. Cummins, B. N. Reinecke, E. Clark, M. K. Sunkara,
T. F. Jaramillo, Nano Lett. 2011, 11, 4168–4175.
23] M. Pourbaix, Atlas of electrochemical equilibria in aqueous solutions,
[
[
31] C. Reichardt, Chem. Rev. 1994, 94, 2319–2358.
32] R. S. Reid, R. J. Clark, E. K. Quagraine, Can. J. Chem. 2007, 85,
1
083–1089.
33] H. E. Gottlieb, V. Kotlyar, A. Nudelman, J. Org. Chem. 1997, 62,
512–7515.
2
nd ed., NACE International, Houston, Texas, 1974.
24] J. Wu, X. Lu, L. Zhang, F. Huang, F. Xu, Eur. J. Inorg. Chem. 2009,
789–2795.
[
7
2
Received: January 24, 2012
Revised: March 18, 2012
Published online: May 15, 2012
25] V. W.-h. Lau, L. G. A. van de Water, A. F. Masters, T. Maschmeyer,
Chem. Eur. J. 2012, 18, 2923–2930.
Chem. Eur. J. 2012, 18, 8230 – 8239
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8239