catalysts

Article
Structure of Nanocrystalline, Partially Disordered
MoS2+δ Derived from HRTEM—An Abundant
Material for Efficient HER Catalysis
Emanuel Ronge 1 , Sonja Hildebrandt 1 , Marie-Luise Grutza 2 , Helmut Klein 3 , Philipp Kurz 2, *
and Christian Jooss 1,4, *
1 Institute of Materials Physics, University of Göttingen, Friedrich-Hund-Platz 1, 37077 Göttingen, Germany;
emanuel.ronge@phys.uni-goettingen.de (E.R.); s.hildebrandt01@stud.uni-goettingen.de (S.H.)
2 Institute for Inorganic and Analytical Chemistry and Freiburg Material Research Center (FMF),
University of Freiburg, Albertstraße 21, 79104 Freiburg, Germany; marie-luise.grutza@ac.uni-freiburg.de
3 GZG Crystallography University of Göttingen, Goldschmidtstr. 1, 37077 Göttingen, Germany;
hklein@uni-goettingen.de
4 International Center for Advanced Studies of Energy Conversion (ICASEC), University of Göttingen,
D-37077 Göttingen, Germany
* Correspondence: philipp.kurz@ac.uni-freiburg.de (P.K.); cjooss@gwdg.de (C.J.);
Tel.: +49-0761-203-6127 (P.K.); +49-551-39-25303 (C.J.)




Received: 5 June 2020; Accepted: 28 July 2020; Published: 1 August 2020


Abstract: Molybdenum sulfides (MoSx , x > 2) are promising catalysts for the hydrogen evolution
reaction (HER) that show high hydrogen evolution rates and potentially represent an abundant
alternative to platinum. However, a complete understanding of the structure of the most active variants
is still lacking. Nanocrystalline MoS2+δ was prepared by a solvothermal method and immobilized
on graphene. The obtained electrodes exhibit stable HER current densities of 3 mA cm−2 at an
overpotential of ~200 mV for at least 7 h. A structural analysis of the material by high-resolution
transmission electron microscopy (HRTEM) show partially disordered nanocrystals of a size between
5–10 nm. Both X-ray and electron diffraction reveal large fluctuations in lattice spacing, where the
average c-axis stacking is increased and the in-plane lattice parameter is locally reduced in comparison
to the layered structure of crystalline MoS2 . A three-dimensional structural model of MoS2+δ could
be derived from the experiments, in which [Mo2 S12 ]2− and [Mo3 S13 ]2− clusters as well as disclinations
represent the typical defects in the ideal MoS2 structure. It is suggested that the partially disordered
nanostructure leads to a high density of coordinatively modified Mo sites with lower Mo–Mo distances
representing the active sites for HER catalysis, and, that these structural features are more important
than the S:Mo ratio for the activity.

Keywords: hydrogen evolution catalysis; molybdenum sulfides; nanocrystalline materials

1. Introduction
The production of hydrogen from renewable energy sources by the splitting of water is a clean
alternative to fossil fuels [1–3]. For a large scale application of water-splitting electrolyzers, abundant,
stable and efficient electrocatalysts for the hydrogen evolution reaction (HER) are needed. However,
the best known materials for this purpose in acidic conditions are noble metals, such as for example,
platinum. Because of the low abundance of noble metals and their resulting high price, the search for
more affordable alternatives is of high interest [4,5].
Due to their relatively low overpotentials and the good availability of Mo, molybdenum sulfides
have gained increasing attention in HER catalyst research during the past few decades [5]. However,

Catalysts 2020, 10, 856; doi:10.3390/catal10080856 www.mdpi.com/journal/catalysts

Catalysts 2020, 10, 856 2 of 16

it is by now accepted that the catalytic activity of crystalline MoS2 in its most common 2H modification
is limited to about 5 mA cm−2 at an overpotential η ~ 500 mV [6] because of the special nature of the
HER active sites, which are mainly present at the edges of the two-dimensional MoS2 planes [7–9].
In contrast, for the 1T polymorph which has a higher HER activity, the crystalline basal planes might
be active sites [9,10]. However, this system is not stable.
In order to improve catalytic activity, sulfur-rich “MoSx ” materials (with x > 2) have been
synthesized and this approach has yielded very promising results [4,11,12]. Some MoSx electrodes
can deliver current densities of 10 mA cm−2 at η ~ 170 mV in 0.5 m H2 SO4 . In special applications,
for example, acidic industrial wastewaters, these catalysts also show a much better long-term stability
than Pt [13]. MoSx materials can be synthesized by solvothermal synthesis [12,14], wet chemical
synthesis [5,15], electrodeposition [4,16], thermal decomposition [17] or chemical oxidation [16].
Typical precursors include ammonium tetrathiomolybdate ((NH4 )2 [MoS4 ]) as a solid [12,14,16],
aqueous solutions containing [MoS4 ]2− [4,16] or ammonium heptamolybdate ((NH4 )6 [Mo7 O24 ]) [5] or
MoO3 reacting with NaS2 [15].
These different synthesis routes for MoSx generally lead to stoichiometries with 2 < x < 4 and
highly disordered, X-ray amorphous structures [4,11,14,18]. X-ray photoemission spectroscopy (XPS)
indicates the presence of a variety of different sulfur species in the form of unsaturated, terminal,
bridging and maybe even apical sulfides and disulfides [4,11,12,14,15].
As results from these studies, two quite different structure models for MoSx have been proposed.
One model suggests a polymer like structure with chains of [Mo2 S9 ]- [11,18–21] or [Mo3 S13 ]-units [16,18,22].
Other models feature a disordered arrangement of [Mo3 S13 ]-clusters [11,12,18,19,23,24]. Sometimes,
crystallization of the amorphous phase to MoS2 nanoparticles during HER is reported [5,25], indicating a
structural affinity to MoS2 .
The polymer model is supported by a publication of Tran et al. [16] using real space scanning
transmission electron microscopy (STEM) imaging of a MoSx sample with a relatively large S:Mo
ratio of about 4. However, the structural analyses of other MoSx with lower sulfur contents (2 <x < 4)
were mostly based on X-ray diffraction (XRD) which does not provide precise structural information
for partially disordered systems. Thus, precise high-resolution real space information about MoSx
structures is of crucial importance for a better understanding of the HER activity and the identification
of active site(s).
The nature of the active sites for disordered MoSx with x > 2 (MoS2+δ ) is controversially discussed
in the literature. Depending on the suggested structural model, different hypotheses exist. If a MoS2 -like
nanostructure arises during HER, there is evidence that terminal disulfides (S2 2− ) at the edges of the
nanocrystals act as active centers [5,25]. However, no crystalline edges are present in amorphous
MoS2+δ but here coordinatively unsaturated molybdenum or sulfur sites are considered to act as HER
active sites [4]. Moreover, density functional theory indicates a higher activity of bridging S2 2− which
would explain the excellent performance of [Mo3 S13 ]2− clusters [11,26]. That S2 2− may be an active site
for MoS2+δ is also supported by XPS studies [12]. Consequently, a better understanding of the actual
HER mechanism of disordered MoSx requires a comprehensive structure model.
Herein, we report on a detailed structural analysis of two solvothermally synthesized MoS2+δ
samples and their hydrogen evolution activity compared to MoS2 , using different electrochemical
measurements (CV, CP and Tafel analyses), high-resolution transmission electron microscopy (HRTEM),
as well as electron- and X-ray diffraction. HRTEM shows that the “X-ray amorphous” structure is in fact
nanocrystalline and features a pronounced disorder within the individual nanocrystals. The detected
fluctuations of in-plane and out of plane lattice parameters measured by X-ray diffraction (XRD),
electron diffraction and HRTEM agree very well. These observations serve as the basis for the
development of a three-dimensional structural model for MoS2+δ , which is qualitatively consistent
with spectroscopic and structural information about MoSx for 2 < x < 4 from literature.
Catalysts 2020, 10, 856 3 of 16

2. Results

2.1. Electrochemical Characterization of MoS2+δ Electrodes

Electrodes were prepared by immobilizing two synthetic MoSx (with x = 2.6 and 3.4, respectively)
and commercially available 2H-MoS2 on graphene. Cyclic voltammograms (CVs), Tafel plots and
chronopotentiometric measurements (CPs) in a strongly acidic electrolyte (0.5 m sulfuric acid, pH 0.3)
were used as descriptors for the electrocatalytic performance (Figure 1). In cyclic voltammograms,
the onset potentials for HER at 3 mA cm−2 were with a value of −190 mV similar for both synthetic
MoSx materials, while MoS2 showed a significantly higher onset potential (−365 mV) and therefore
lower catalytic activity.
Another common descriptor for the HER activity is the overpotential (η) needed to reach current
densities of 10 mA cm−2 in the CVs (Table 1). MoS2.6 showed the best performance with an overpotential
of η = 235 mV, followed by MoS3.4 with η = 245 mV and MoS2 with η = 450 mV. This is in agreement
with the literature [4,6], where it is also described that disordered MoSx are far more active HER
catalysts than crystalline MoS2 . For all of these values, one has to keep in mind that the current
densities are derived for the geometric area of the electrodes. However, the difference in activity
between the prepared samples could also lie in their different stoichiometric compositions. Assuming
a molybdenum-based proton reduction mechanism as postulated for example, by Tran et al. [16],
defective structures and molybdenum-rich materials would be in favor of high proton reduction activity.
Indeed, when the detected currents are normalized to the amount of molybdenum on the electrodes,
MoS2.6 and MoS3.4 show nearly identical overpotentials of η = 250 and 255 mV at 10 mA µmol(Mo) −1 ,
while MoS2 is not able to reach this current (see Figure S7). Concluding from these results, a high sulfur
content seems to be much less important for the HER catalysis rate than the existence of a disordered
structure for 2 < x < 3.5.
In addition, the particle sizes of the synthesized materials differ with (0.18 ± 0.06) µm significantly
from the commercial MoS2 with particle sizes of (0.8 ± 0.9) µm showing a much broader dispersity
of the particles (see Figure S6). Hence, the prepared MoSx both contain more accessible active sites
than MoS2 where the active sites are believed to be the structural sulfur-rich edges [7]. However,
the different particle sizes and shapes result in an increase of the MoS2+δ surface by a factor of only
about 1.9 compared to MoS2 (see ESI for details). Assuming the same degree of porosity, this indicates
that the higher activity of the MoS2+δ electrodes compared to the MoS2 electrode cannot be explained
by morphology effects alone.
Tafel slopes were calculated from chronoamperometric “staircase measurements” (see materials
and methods) and values of 100 mV dec−1 for MoS2.6 , 90 mV dec−1 for MoS3.4 and 160 mV dec−1 for
MoS2 were obtained for the same catalyst loadings. According to these values, the rate limiting step
for MoS2 might be the adsorption of an H atom (the Volmer reaction), while the results for MoS2.6 and
MoS3.4 do not give clear evidence on the rate limiting step as the values lie between the boundary
values of 40 mV dec−1 for the Heyrovsky reaction (reductive desorption) and the Volmer reaction
(>120 mV dec−1 ) [13].
All of the measured values for overpotentials and Tafel slopes of the MoSx studied here cannot
compete with the best MoSx that are known today (Table 1), which is hardly surprising given the
fact that both material synthesis and electrode fabrication were not optimized. However—and most
important for the following detailed structural characterization—both MoSx samples show very
respectable electrocatalytic HER activity while the MoS2 reference does not.
Catalysts 2020, 10, 856 4 of 16
Catalysts 2020, 10, x FOR PEER REVIEW 4 of 16

Figure 1. Electrochemical measurements for the two MoSx and MoS2 (immobilized on graphite) in
Figure 1. Electrochemical measurements for the two MoSx and MoS2 (immobilized on graphite) in sulfuric
sulfuric acid (0.5 M, pH 0.3). (A): cyclic voltammetry (cycle 6, scan rate 20 mV s−1); (B):
acid (0.5 m, pH 0.3). (A): cyclic voltammetry (cycle 6, scan
−2
rate 20 mV s−1 ); (B): chronopotentiometry at a
chronopotentiometry at a current density of 3 mA cm , the first 2 min are cut off; (C): Tafel analysis.
current density of 3 mA cm−2 , the first 2 min are cut off; (C): Tafel analysis. See materials and methods
See materials and methods for experimental details.
for experimental details.
The long-term stability of the electrodes was tested in CP measurements (Figure 1B). Herein,
The long-term stability of the electrodes was tested in CP measurements (Figure 1B). Herein,
both synthetic MoSx materials clearly outperformed crystalline molybdenum disulfide over a period
bothofsynthetic
7 h. For aMoS x materials
set current clearly
density outperformed
of 3 mA cm−2, at the crystalline molybdenum
end of the experiments MoSdisulfide over a period of
2.6 and MoS3.4 showed
7 h. overpotentials
For a set current density −2
of ~210 mV ofand3 mA
~195cm mV, ,respectively,
at the end of inthe experiments
comparison MoS
to ~430 mV andMoS
2.6 for MoS 2. 3.4
showed
In these
overpotentials
measurements, of ~210
both mV and ~195
synthetic MoSx mV, respectively,
showed extremely in comparison
stable performances to ~430
withmV for MoS
negligible 2 . In these
increases
measurements, both synthetic
of the overpotentials. MoSx showed
The difference extremely
in the catalytic stable performances
activity of MoSx might with
be negligible
due to some increases
MoS2 of
the overpotentials.
particles as the The difference
powder in the catalytic
diffractogram shows aactivity
sharp of MoS
peak might
atx (0 be due
0 2) for MoSto2.6.some
Over MoS
time,2 particles
MoS2
showed
as the powder a slight increase ofshows
diffractogram the HER overpotential,
a sharp peak at (0which
0 2) might
for MoS be2.6
due to a reduced
. Over time, MoS electric resistivity
2 showed a slight
of theofcatalyst
increase the HER layer related to a shrinking
overpotential, which mightthickness
be dueand/or
to a structural rearrangements
reduced electric resistivity ofof
thethesurface
catalyst
layerleading
related to to
a higher amountthickness
a shrinking of active sights.
and/or structural rearrangements of the surface leading to a
higher amount of active sights.
Catalysts 2020, 10, x FOR PEER REVIEW 5 of 16
Catalysts 2020, 10, 856 5 of 16
Table 1. Tafel slopes and overpotentials at 10 mA cm−2 of MoSx and MoS2.

Tafel
Table 1. Tafel slopes and Slope ηat@1010mA
overpotentials mAcmcm η @ 10
−2 of MoS mA μmol(Mo)
−2 −1
x and MoS2 .
[mV dec ] −1 [mV] [mV]
MoS2.6 100−1 ]
Tafel Slope [mV dec η @ 10 mA cm−2 [mV]
235 η @ 10 mA
250µmol(Mo)−1 [mV]
MoS2.6 MoS3.4 100 90 245
235 255 250
MoS3.4 MoS2 90 160 245
450 - 255
MoS2 160 450 -
MoS2 [27] 160 - -
MoS2 [27] 160 - -
MoS2 MoS2
59 59 104
104 - -
(step-edged stacks) [28]
(step-edged stacks) [28]
1T MoS21T(porous) [29]
MoS2 (porous) [29] 43 43 153
153 - -
MoS2+x [4] 40 160 -
MoS2+x [4] 40 160 -
MoS3.5 [5] ≈60 ≈200 -
MoS3.5 [5] ≈60 ≈200 -

2.2. Electron
2.2. Electron andand X-ray
X-ray Diffraction
Diffraction Analysis
Analysis of of the
the MoS
MoS2+δ Structure
2+δStructure

For the
For thedetermination
determinationofof thethe
structure of MoS
structure of 2+δ
MoS , XRD measurements were carried out. The powder
2+δ, XRD measurements were carried out. The
diffractograms for the two
powder diffractograms for the two MoS samples (MoS
2+δMoS2+δ samples (MoS 2.6 & MoS 3.4 ) from
2.6 & MoS
different batches are given in
3.4) from different batches are given
Figure
in Figure2. 2.
ForForcomparison,
comparison,aapowder powder XRD XRD of the crystalline
of the crystallineMoS MoS2 2sample
samplewas was measured
measured as as well.
well. A
A Rietveld refinement of the
Rietveld refinement of the MoS2 data MoS data was conducted, which yielded results
2 was conducted, which yielded results in good agreement with in good agreement
withstructural
the the structural
model model of Wildervanck
of Wildervanck et [30].
et al. al. [30].
TheThe results
results areare shown
shown ininthetheESI
ESIininFigure
FigureS1 S1 and
and
Tables S2
Tables and S3.
S2–S3. In contrast
In contrast to to
MoS MoS 2 , the
2, the XRDXRD reflexesofofMoS
reflexes MoS2+δ 2+δ
arearestrongly
stronglybroadened.
broadened.Both Bothsamples
samples
show a very wide peak at about 8 Å which deviates about 1–2 Å
show a very wide peak at about 8 Å which deviates about 1–2 Å from the literature value for MoSfrom the literature value for MoS22..
Besides this
Besides this difference,
difference, thethecenters
centersof ofthe
theother
otherreflexes
reflexesofofthe theMoS
MoS 2+δ
2+δ
XRD
XRD pattern
pattern fit fit
thethe positions
positions of
of the MoS 2 data (see Table S1). The main difference among the MoS
the MoS2 data (see Table S1). The main difference among the MoS2+δ samples themselves is the sharp 2+δ samples themselves is the
sharp double
double peak of peak of sample
sample MoS2.6MoS 2.6 , which
, which reflectsreflects the presence
the presence of some of larger
some larger
more more MoS2grains.
MoS2-like -like grains.
This
This might indicate a convergence of the MoS 2+δ structure towards MoS
might indicate a convergence of the MoS2+δ structure towards MoS2 for decreasing S : Mo ratios. The 2 for decreasing S:Mo ratios.
The presence
presence of only
of only one one
sharpsharp
MoSMoS 2 reflex
2 reflex (002) (002)
in MoSin MoS 2.6 can
2.6 can
be explained
be explained by aby a preferred
preferred orientation
orientation of
of MoS 2 –like crystals due to their plate like morphology
MoS2 –like crystals due to their plate like morphology (see Figure S6A). (see Figure S6A).

Figure
Figure 2. Powder
Powder X-ray
X-ray diffraction
diffraction (XRD)
(XRD) pattern
pattern ofof MoS
MoS22 and
and the
the nanocrystalline,
nanocrystalline, partially
partially
disordered samples MoS
disordered samples 2.6 and MoS3.4
MoS2.6 3.4.. The
The indicated
indicated diffraction maximaare
diffraction maxima aredetermined
determinedbybyfitting
fittingofofa
agaussian
gaussianfunction.
function.
Catalysts 2020, 10, 856 6 of 16
Catalysts 2020, 10, x FOR PEER REVIEW 6 of 16

In
In addition
addition to the XRDXRD analysis,
analysis,selected
selectedarea
areaelectron
electron diffraction
diffraction (SAD)
(SAD) waswas carried
carried out.out.
The
The electron diffraction patterns of the two samples
electron diffraction patterns of the two samples MoS2.6 and MoS MoS and MoS
2.6 3.4 are shown
3.4 are shown in Figure S2
in Figure S2 and consist
and consist
of only of only
of rings, whileof no
rings,
spotswhile no spots
are visible. are visible.
By circular By circular
integration of theintegration
intensities, of
thethe intensities,
profiles shown
the
in profiles
Figure 3shown in Figure 3Three
were obtained. werevery
obtained.
broadThree very
reflexes broad
are reflexes
visible in theare visible in
intensity the intensity
profiles of both
profiles
samples of while
both samples while
the values fortheMoS
values forshifted
2.6 are MoS2.6 toareslightly
shifted to slightly
lower lowerNevertheless,
values. values. Nevertheless,
both are
both are consistent
consistent with thewith
MoS the MoS2 structure
2 structure withinwithin the measurement
the measurement accuracy.
accuracy. A detailed
A detailed discussion
discussion of
of the
the peak positions of MoS
peak positions of MoS2+δ 2+δ compared to MoS
compared to MoS2 [30,31] 2 [30,31] can be found in the supplement (see
can be found in the supplement (see Table S4). Table S4).

Figure3.3.Intensity
Figure Intensityprofiles
profilesofofthe
therepresentative
representativeelectron
electrondiffraction
diffraction patterns
patterns ofof
thethe two
two MoSMoS2+δbatches
batches
2+δ
investigated in this study (MoS 2.6 in dark blue & MoS3.4 in light blue). Vertical lines indicate the
investigated in this study (MoS2.6 in dark blue & MoS3.4 in light blue). Vertical lines indicate the
positionsofofdiffraction
positions diffraction maxima
maxima andandthe the marked
marked areasareas indicate
indicate the width
the width of theof the reflexes.
reflexes. Note
Note the the
reflex
atreflex
≈4.53atÅ≈4.53 Å islikely
is most mostanlikely an artefact
artefact duebeam
due to the to thestopper.
beam stopper.

Theresults
The resultsofofthe
theX-ray
X-rayandandelectron
electrondiffraction
diffractionexperiments
experimentsindicate
indicatethat
thatthe
thestructure
structureofofthethe
MoS
MoS2+δ2+δ samples is comparable and compatible with a disordered MoS structure independent
compatible with a disordered MoS2 structure independent of their
2 of their
S : Mo
S:Mo ratio.
ratio. The
The shift
shift of of
thethe first
first peak
peak of of
thethe MoS
MoS samplescompared
2+δ2+δsamples comparedtotothethe(002)
(002)reflex
reflexofofMoS
MoS 22
towardslower
towards lowerangles
anglesandandthe
thevariation
variationofofthe
thepeak
peakcenter
centerreflects
reflectsan
anenlargement
enlargementand andfluctuations
fluctuationsofof
thelattice
the latticeparameter
parameterinin[001]
[001]direction.
direction.

2.3.
2.3.HRTEM
HRTEMAnalysis
Analysisofofthe
theMicrostructure
MicrostructureofofMoS
MoS2+δ
2+δ

To
Togain
gainaadeeper
deeperinsight
insightinto
intothe
thestructure,
structure,HRTEM
HRTEManalysis
analysiswas
wasperformed.
performed.During
DuringthetheHRTEM
HRTEM
analysis,
analysis,no
nosignificant
significantchanges
changesto tothe
thecrystal
crystalstructure
structurewaswasobserved
observedover
overtime
timeatataabeam
beamdose
doserate
rate
ofof10,000
10,000toto 64,700 /(Å22 s).
64,700 ee−−/(Å s). Thus,
Thus,we wefind
findthat
thatMoS
MoS is
2+δ
2+δ is stable
stable under
under thethe electron
electron beam
beam which
which is a
isrequirement
a requirement for a reliable HRTEM analysis. The HRTEM images
for a reliable HRTEM analysis. The HRTEM images of MoS2+δ 2+δ of MoS reveal a bended
reveal a bended and
and partially
partially disordered
disordered crystal
crystal layer
layer stacking.
stacking. Coherencelength
Coherence lengthofofordered
ordered stacking
stacking along thethe (001)
(001)
direction is about 5 nm and ordering within the planes is reaching up to 10 nm. The in-plane ordering
Catalysts 2020, 10, 856 7 of 16

Catalysts 2020, 10, x FOR PEER REVIEW 7 of 16

direction is about 5 nm and ordering within the planes is reaching up to 10 nm. The in-plane ordering
can also be observed along [001] zone axis: A HRTEM image of such a crystal plate is presented in
can also be observed along [001] zone axis: A HRTEM image of such a crystal plate is presented in
Figure 4.
Figure 4.

Figure 4. High-resolution transmission electron microscopy (HRTEM) image of sample MoS3.4

Figure 4. High-resolution transmission electron microscopy (HRTEM) image of sample MoS3.4 showing
showing the in-plane ordering. (A): Overview of the edge of a particle; (B): zoom in the area marked
the in-plane ordering. (A): Overview of the edge of a particle; (B): zoom in the area marked with a red
with a red square in A; (C): Fast Fourier Transformation (FFT) of the area shown in B.
square in A; (C): Fast Fourier Transformation (FFT) of the area shown in B.

Fast Fourier
Fast Fourier Transformation
Transformation (FFT) (FFT) shows
shows the the typical
typical hexagonal
hexagonal symmetry
symmetry of of the
the (100)
(100) lattice
lattice
planes of
planes of MoS
MoS2.. The discovery of this ordering with a hexagonal lattice symmetry of atomic positions
2 The discovery of this ordering with a hexagonal lattice symmetry of atomic positions
within the layers as
within the layers as well aswell as an
an ordering
ordering of of the
the layers
layers along
along thethe (001)
(001) direction
direction support
support our
our conclusion
conclusion
that MoS 2+δ and MoS2 show a close structural affinity. The correlation length along (001) is ξc ≈ 5 nm,
that MoS2+δ and MoS2 show a close structural affinity. The correlation length along (001) is ξc ≈ 5 nm,
while from
while from thethe in-plane
in-planeordering
orderingan anin-plane
in-planecorrelation
correlation length
length ξabξ≈ 4 ≈nm
of of 4 nmcan can
be determined.
be determined.The
ab
reason for the deviation from the prior determined length of MoS 2+δ planes of 10 nm is the bending
The reason for the deviation from the prior determined length of MoS2+δ planes of 10 nm is the bending
of the
of thecrystals
crystalsplanes
planeswhich
which disturbs
disturbs thethe
phasephase contrast
contrast in theintopthedown
top down
view. view.
Such aSuch a correlation
correlation length
length in the order of 1–5 nm is also called medium range order and can be
in the order of 1–5 nm is also called medium range order and can be found in nanocrystalline [32] as wellfound in nanocrystalline
[32]
as inas well as in amorphous
amorphous systems [33].systems
From the [33].
XRDFrom the XRD measurements,
measurements, a correlationa length
correlation length
of (1.22 of (1.22
± 0.10) nm
± 0.10)
can nm can bewith
be calculated calculated with the Scherrer-Formula.
the Scherrer-Formula. It is well known It is well
that theknown that the Scherrer-Formula
Scherrer-Formula underestimates
underestimates
the grain size for the grain size
partially for partially
disordered disordered since
nanoparticles, nanoparticles,
it measures sincetheit coherence
measures the coherence
length of the
length of
lattice [34]. the lattice [34].
AA more
more detailed
detailed analysis
analysis ofof the
the HRTEM
HRTEMimages imagesfor forMoSMoS2+δ was carried out using FFT and its
2+δ was carried out using FFT and its
results are exemplified in Figure 5 C. The lattice distance d (100) exhibits a general trend of being locally
results are exemplified in Figure 5 C. The lattice distance d(100) exhibits a general trend of being locally
reduced compared
reduced compared to to the
the literature
literature values
values of ofMoS
MoS2 (d (d{100} == 2.7368
2.7368 Å Å [30])
[30]) with
with an average value
an average value ofof
2 {100}
d(100) ==(2.69
d (2.69±±0.21)
0.21)ÅÅand andlocal
localvariations
variationsfrom from2.45
2.45ÅÅtoto2.79 2.79ÅÅ(see (seealso
alsoTable
TableS5S5ininthe
theESI
ESIforfora
(100)
summary). Within the measurement accuracy and statistics no significant
a summary). Within the measurement accuracy and statistics no significant difference between the difference between the two
S : Mo
two S:Mo ratios
ratioscan
canbebeobserved.
observed.Both Bothsamples
samples(MoS (MoS2.62.6&&MoS ) )show
MoS3.43.4 showsimilar
similar reduction
reduction and local
and local
variations in the Mo–Mo
variations in the Mo–Mo distance. distance.
Catalysts 2020, 10, 856 8 of 16

Catalysts 2020, 10, x FOR PEER REVIEW 8 of 16

Figure5.5.HRTEM
Figure HRTEManalysis analysisofofnanocrystalline
nanocrystallineMoS MoS 2+δ.. (A):
2+δ (A): HRTEM
HRTEMimagesimagesofofMoS MoS 2.6;; (B):
2.6 (B):HRTEM
HRTEM
images
imagesofofMoS
MoS3.43.4;; (C):
(C): representative
representativeFFT FFTintensity
intensityprofiles.
profiles.TheTheFFT
FFTofofMoS
MoS22isis taken
taken from
fromFigure
FigureS4A,
S4A,
and
andthe
the other underlying FFTs
other underlying FFTs cancanbe befound
foundininFigure
Figure S5;S5;
TheThe in-plane
in-plane lattice
lattice spacing
spacing of MoSof MoS
2 d1002=

100 = Å
d2.737 2.737 Å (Wildervanck
(Wildervanck et al.is[30])
et al. [30]) is indicated.
indicated. The region
The region of interest
of interest of the of
FFT theplots
FFT isplots
markedis marked
with a
with a rectangular in subfigure A and B with the
rectangular in subfigure A and B with the corresponding color. corresponding color.

AAclose
closeup upof ofaacross
crosssection
sectionimage
imageof ofthe
thecrystal
crystallayers
layersisisshown
shownin inFigure
Figure6A.6A.The
Thelayers
layersarearenot
not
perfectly flat and parallel which leads to a variation in layer distances. On average, the
perfectly flat and parallel which leads to a variation in layer distances. On average, the lattice stacking lattice stacking
distance
distancedd(001) of MoS 2+δ varies
(001) of MoS2+δ
varies from
from 11.15
11.15 ÅÅ to
to 15.93
15.93 Å.
Å. The
The overview
overviewin inTable
TableS5S5indicates
indicatesthat
thatd(001)
d(001)
isisgenerally enlarged when compared to the literature value of
generally enlarged when compared to the literature value of MoS22 (d{001} MoS (d = 12.294 Å) [30] for both
{001} = 12.294 Å) [30] for both S
S:Mo
: Mo ratios,
ratios, which
which is is in
in good
good agreement
agreementwithwithXRDXRDanalysis.
analysis. Next
Next toto dislocations
dislocationsalso alsodisclinations
disclinationsareare
visible in Figure 6B (see also Figure S4B). This is quite unusual for solid state
visible in Figure 6B (see also Figure S4B). This is quite unusual for solid state materials materials but, for example,
but, for
has been observed
example, has beenin fullerene-like
observed variants ofvariants
in fullerene-like MoS2 byofSrolovitz
MoS2 byetSrolovitz
al. [35]. et al. [35].
Catalysts 2020, 10, 856 9 of 16

Catalysts 2020, 10, x FOR PEER REVIEW 9 of 16

Figure 6. Close up of the HRTEM image from the left side of Figure 5 B, showing the MoS3.4 lattice
Figure 6. Close up of the HRTEM image from the left side of Figure 5 B, showing the MoS3.4 lattice
planes in cross section view. (A): Position of the intensity profile with a width of 10 px and direction
planes in cross section view. (A): Position of the intensity profile with a width of 10 px and direction
marked by arrows. (B): Disclinations formed by merging MoSx layers are highlighted by cyan lines.
marked by arrows. (B): Disclinations formed by merging MoSx layers are highlighted by cyan lines.
(C): Intensityprofiles
(C): Intensity profilesindicated
indicatedin
inAAand
andexemplary
exemplaryvariations
variationsin
inlattice
latticedistance.
distance.For
Forthe
thefull
fullanalysis
analysis
seeTable
see Tables 2 and
2 and S5. S5.
Table

In order
In order toto provide
provide accurate
accurate calibration
calibration of of the
the HRTEM
HRTEM images,images, all all images
images werewere calibrated
calibrated by
by
measurements of gold particles. In addition, the crystalline MoS sample
measurements of gold particles. In addition, the crystalline MoS2 sample was also analyzed by HRTEM,
2 was also analyzed by
HRTEM,
as shown in asFigure
shownS4. inBoth
Figure S4. Both
average average
lattice latticedparameters
parameters (100) and d(001) d(100)
areand d(001) are in
in agreement to agreement to
the literature
the literature
values values
[30] within the[30] within(see
accuracy the experimental)
accuracy (see experimental)
as well as to our as XRD
well asdata.to our XRD data.
Insome
In somecases,
cases, a recrystallization
a recrystallization of amorphous
of amorphous MoSxMoS underx under the electron
the electron beam was beam was observed
observed [36,37].
[36,37].
In In addition,
addition, Xi et al.Xi[37]
et al. [37] report
report on theon the transformation
transformation of amorphous
of amorphous MoSMoS x x after
after 2 h 2ofh of
HERHERto to
a
nanocrystalline material which is stable under the electron beam. The crystal diameter is comparable tois
a nanocrystalline material which is stable under the electron beam. The crystal diameter
comparable
the ones shown to the ones shown
in Figures 4 andin 5. Figures
However, 4 and
they5.observe
However, they degree
a lower observeofastacking
lower degree of stacking
of crystal planes
of crystal to
compared planes
Figure compared to Figure
6, indicating a higher 6, degree
indicating a higher
of order in ourdegree
system. of The
order in our system.
transformation of The
the
transformation of the amorphous to nanocrystalline MoS in Xi et al. [37]
amorphous to nanocrystalline MoSx in Xi et al. [37] was also accompanied by an increase in hydrogen
x was also accompanied by
an increase indicating
production in hydrogen production
that indicating that
the nanocrystalline theisnanocrystalline
phase the more active phase
andisstable
the more active and
configuration.
stable configuration.
Together with our results Together with our
this suggests that results
our twothis
MoSsuggests
2+δ
that
samples our
both two
representMoS a samples
thermodynamic
2+δ both
represent a thermodynamic
stable and active form of MoSx . stable and active form of MoS x .

3.
3. Structure
Structure Model
Model
The
Thecombined
combined application
applicationof electron and X-ray
of electron andprobes
X-rayreveal that MoS
probes 2+δ exhibits
reveal that MoS a nanocrystalline,
2+δ exhibits a

partially disordered
nanocrystalline, MoS2 -like
partially structure
disordered MoS with
2-likea locally reduced
structure with alattice
locally parameter a and an
reduced lattice enlargeda
parameter
mean
and an enlargedc,mean
parameter with both showing
parameter strong
c, with fluctuations.
both This follows
showing strong from theThis
fluctuations. observed
follows hexagonal
from the
in-plane
observedsymmetry
hexagonal of in-plane
the crystallites,
symmetry the c-axis
of thestacking of planes
crystallites, and the
the c-axis measured
stacking fluctuations
of planes of
and the
the lattice distances
measured fluctuationsrevealed by HRTEM
of the lattice as well
distances as diffraction
revealed by HRTEM techniques.
as well asAdiffraction
suggestedtechniques.
structural
model for MoSstructural
A suggested 2+δ consistent
model with
for these
MoS2+δ results is depicted
consistent in Figure
with these results7. Table 2 compares
is depicted the measured
in Figure 7. Table 2
mean lattice
compares spacing
the in [001]
measured meanand [100]spacing
lattice directions as well
in [001] andas [100]
their directions
fluctuations aswith
well the literature
as their values
fluctuations
with
for MoSthe2 .literature values
The results for dfor
100MoS
from 2 . The
TEM results
are in for
goodd from
agreement
100 TEM are
with inthegood agreement
X-ray diffraction with the X-
analysis.
ray diffraction analysis. In addition to the small grain sizes of a few nm, the strong disorder within
In addition to the small grain sizes of a few nm, the strong disorder within the nanocrystals is consistent
the nanocrystals
with is consistent
the broad reflexes in XRDwith and theSAD. broad reflexes in XRD and SAD.
Catalysts 2020, 10, 856 10 of 16

Catalysts 2020, 10, x FOR PEER REVIEW 10 of 16

Table 2. Comparison of the (100) and (001) lattice distances of MoS2+δ found by transmission electron
Table 2. Comparison
microscopy (TEM) and of thewith
XRD (100)literature
and (001)values
latticefor
distances of MoS
MoS2 [30]. found
The2+δXRD by transmission
uncertainty electron
is determined
microscopy (TEM) and XRD with
by the full width at half maximum (FWHM).literature values for MoS 2 [30]. The XRD uncertainty is determined

by the full width at half maximum (FWHM).

Literature [30] HRTEM MoS2+δ
Literature [30] HRTEM MoS2+δ XRD
MoS2 Mean Min Max XRD
MoS2 Mean Min Max +0.16
d(100) [Å] 2.7368 2.69 ± 0.21 2.45 ± 0.18 2.79 ± 0.03 2.69−0.39
+ 10
+0.16
(100) [Å]
d(001)d[Å] 2.7368
12.294 2.69±±0.8
12.9 0.21 11.15
2.45 ±±0.18 2.79 ±±0.03
0.19 15.93 0.61 2.6915-0.39
−4
+10
d(001) [Å] 12.294 12.9 ± 0.8 11.15 ± 0.19 15.93 ± 0.61 15-4

In agreement with the stoichiometry of MoS2+δ , the excess sulfur must be present in the
In agreement with the stoichiometry of MoS2+δ, the excess sulfur must be present in the form of
form of unsaturated, terminal, bridging and possibly apical sulfides and disulfides, as visible by
unsaturated, terminal, bridging and possibly apical sulfides and disulfides, as visible by XPS
XPS [4,11,12,14,15]. This modification results in structural changes relative to the 2H-MoS2 structure.
[4,11,12,14,15]. This modification results in structural changes relative to the 2H-MoS2 structure. Our
Our model suggests that some parts of the structure show similarities to the atomic arrangements
model suggests that some parts of
2− and the structure show similarities to the atomic arrangements found
found in [Mo 2 S12 ] [Mo3 S13 ]2− clusters. Like MoS2+δ these clusters have also a reduced Mo–Mo
in [Mo2S12]2− and [Mo3S13]2− clusters. Like MoS2+δ these clusters have also a reduced Mo–Mo distance
distance compared to MoS2 (see Table 3). The disorder in the in-plane lattice distance of MoS2+δ thus
compared to MoS2 (see Table 3). The disorder in the in-plane lattice distance of MoS2+δ thus reflects
reflects local sulfur-rich disorder in the form of cluster like structural units which are incorporated into
local sulfur-rich disorder in the form of cluster like structural units which are incorporated into the
the MoS2 nanocrystals.
MoS2 nanocrystals.

Figure 7. Schematic illustration of the suggested structure model for MoS2+δ . The indicated a-, b- and
Figure 7. Schematic illustration of the suggested structure model for MoS2+δ. The indicated a-, b- and
c-axis represent the unit cell and lattice parameters of MoS2 . (A): plane view; (B): out of plane view.
c-axis represent the unit cell and lattice parameters of MoS2. (A): plane view; (B): out of plane view.

Table 3. Mo-Mo distance of MoS2+δ compared with MoS2 [30] and the cluster anions [Mo2S12]2− and
[Mo3S13]2− [18].
Catalysts 2020, 10, 856 11 of 16

Table 3. Mo-Mo distance of MoS2+δ compared with MoS2 [30] and the cluster anions [Mo2 S12 ]2− and
[Mo3 S13 ]2− [18].

MoS2+δ MoS2 [Mo2 S12 ]2− [Mo3 S13 ]2−

Mean Min Max [30] [18] [18]
dMo-Mo [Å] 3.11 ± 0.24 2.83 ± 0.21 3.22 ± 0.04 3.16 ≈2.8 ≈2.7

Compared to the [100] direction, the variations in lattice spacings in the [001] direction are larger
and the XRD analysis indicates an overall increased distance between the MoS2 layers, which are only
bound to each other by weak Van der Waals interactions. The previously described in-plane variations
and cluster-like disorder also can lead to local alternations in out of plane sulfur positions which might
affect the Van der Waals bonding distance and thus induce a varying layer spacing. In particular,
the alternation of the layer spacing at the nanocrystallite edges, as well as at the disclinations, is very
large. In addition, external stress from boundaries to other neighboring crystals can induce further
lattice spacing modulations.
Our suggested model is in qualitative agreement with literature results for disordered MoSx with
2 < x < 4. MoS2+δ is generally highly disordered. The in-plane structure of the detected nanocrystals
partially features a hexagonal symmetry like MoS2 . The local defect structures show similarities to
[Mo2 S12 ]2− and [Mo3 S13 ]2− clusters, which correlates well with the increased sulfur content compared
to MoS2 .
Hinnemann et al. [7] studied MoS2 nanoparticles with approximately 4 nm in diameter and
1 nm in apparent height on graphite and stated that only the edges of MoS2 are interesting in the
context of HER, as the basal plane of MoS2 is catalytically inactive. Our structure model for MoS2+δ
strongly features frayed edges similar to nanocrystalline MoS2 and due to the high defect concentration,
coordinatively modified Mo sites also appear within the basal planes. Consequently, we expect that
some of the active sites are similar to the report of Hinnemann et al. [7]. But in addition, cluster-like
structures appear within the lattice planes as well as at their edges and in the disclinations. These planar
structures exhibit a partial stacking and ordering along the c-axis, also establishing a similarity to
the MoS2 crystal, however, with increased lattice parameters c due to small crystal sizes as well as
disorder in the in-plane structure. Typically, XPS for MoSx with 2 < x < 4 indicates bridging and
terminal disulfides as well as unsaturated molybdenum and sulfur ions [4,11,12,14,15]. The structure
observed here featuring disordered nanocrystals with a size of a few nanometers can thus explain a
high density of catalytically active sites which are present at the defective nanocrystal planes as well as
at their edges. This, in consequence, could very well explain the much higher HER activity of MoS2+δ
compared to MoS2 .
Wu et al. [38] also report a reduced Mo–Mo distance of 2.778 Å for their amorphous MoSx ,
which is in good agreement with this work (smallest Mo–Mo distance (2.81 ± 0.2) Å) and suggest it as
a key feature for the higher activity as the electronic structure gets even more similar to the clusters.
In addition, no influence of the sulfur dimer content on the activity was observed by Wu et al. [38].
This is in good agreement with this work, where the catalytic activity tends to correlate with Mo content.
Considering also the high catalytic activity of disordered nanocrystalline MoSx with x < 2 reported by
Xi et al. [37] this might indicate the larger impact of the presence of coordinately modified Mo sites and
reduced lattice parameter rather than the S:Mo ratio. In addition, Ying et al. [29] reports on improved
catalytic activity by increasing the concentration of sulfur vacancies. This supports to allocate the
active site to the Mo edges. In addition, the structure model for nanocrystalline, partially disordered
MoS2+δ is based on stoichiometry compensating defects that change the Mo coordination and Mo–Mo
bonding distance. Thus, both the S:Mo ratio and the processing induced microstructure influence the
crystal structure.
Catalysts 2020, 10, 856 12 of 16

4. Materials and Methods

All chemicals were purchased commercially and, if not stated otherwise, used without further
purification. Deionized water (R = 18.2 MΩ) from an Elga Veolia PURELAB flex 4 water purification
system was used for all experiments. Crystalline MoS2 was purchased from Sigma-Aldrich.

4.1. Synthesis of MoSx

Ammonium tetrathiomolybdate (NH4 )2 [MoS4 ] was prepared following a literature procedure by
McDonald et al. [39]. The very high purity of the (NH4 )2 [MoS4 ] was confirmed by mass and Raman
spectroscopy as well as XRD. To obtain amorphous molybdenum sulfide, a slightly modified method
from the route of Li et al. [40] was used: (NH4 )2 [MoS4 ] (100 mg) was dissolved in water and hydrazine
(N2 H4 ·H2 O, 0.3 mL) was added. The mixture was transferred into a steal autoclave with a Teflon
inlet (45 mL), heated up to 200 ◦ C for 12 h and then allowed to cool down to room temperature.
The reaction mixture was centrifuged (10 min @ 5000 rpm), washed with THF (3 × 25 mL) as well as
water (2 × 30 mL) and freeze-dried. As comparably high amounts of MoSx were required for the XRD
and TEM analyses, two batches of the product (MoS2.6 and MoS3.4 ) were synthesized due to the rather
small volume of the available steel autoclave. Both MoSx batches were obtained as black powders in
very similar yields of ~60 mg each.

4.2. Stoichiometry Determination

An Analytik Jena novAA® 350 flame atomic absorption spectrometer (F-AAS) (Analytik Jena, Jena,
Germany) was used to determine S:Mo ratios. The calibration was performed using an ammonium
heptamolybdate solution diluted by 0.03% v/v aqua regia. Prior to analysis, the samples (5 mg) were
completely dissolved in aqua regia (5 mL) to oxidize molybdenum to Mo(VI). This solution was diluted
to 200 mL and used without further dilution. The measurements were conducted five times for each
sample and the mean value taken. The S:Mo ratios 2.6 ± 0.13 and 3.4 ± 0.17 were calculated from
the initial sample weight and the F-AAS results presuming the absence of any other elements than
molybdenum and sulfur. Another reason for the divergent composition could lie in the synthesis
process. The addition of the reductant hydrazine is not performed in a controlled manner; neither is
the solution stirred during the synthesis. This and the fact that the reaction mixture possibly contains
high amounts of different sulfur species like sulfides, disulfides as well as hydrazine, ammonia and
water in variable proportions could explain the differing S:Mo ratios.
X-ray fluorescence (XRF) (M4 Tornado from Bruker Corporation, Billerica, Massachusetts, USA)
measurements calibrated with MoS2 resulted in a similar S:Mo ratio of about 3. However, due to
a strong peak overlap between molybdenum and sulfur, the resulting S:Mo ratio uncertainties are
quite high when using XRF, energy-dispersive X-ray spectroscopy (EDS) (INCA detector from Oxford
Instruments, Abingdon, England) or wavelength-dispersive X-ray spectroscopy (WDS) (JXA-8900 RL
from JEOL, Akishima, Tokyo, Japan), which makes AAS the most reliable method in this case.

4.3. Electrochemistry
For the electrochemical measurements, a Princeton Applied Research Versa Stat 4 potentiostat
(AMETEK Princeton Applied Research, Oak Ridge, TN, USA) was used. All measurements were
performed in a three-electrodes setup using MoS2 resp. MoSx on graphite sheets as working electrodes
(WE), platinum as counter electrode (CE) in a separate compartment with a glass frit and an Ag/AgCl
electrode as counter electrode (3 m KCl, RE). Sulfuric acid (0.5 m, pH 0.3) served as electrolyte. Cyclic
voltammograms (CV) were recorded in a range of 0.1(−0.5) VRHE with a sweep rate of 20 mV s−1 .
Tafel slopes were determined from the sixth cycle (first half) of the CV. All electrochemical measurements
were iR-corrected at 85%.
Catalysts 2020, 10, 856 13 of 16

4.4. Electrode Preparation

To eliminate impurities from the graphite sheet, the blank electrodes were cleaned with
water, isopropanol and ethanol. The electrodes were prepared following procedures published
by Cui et al. [41] 3 mg of the catalyst and 6 µL of Nafion solution (5 wt.%) were dispersed in 0.6 mL
water/ethanol (49:50, v:v) and treated with ultrasound for 30 min. 48 µL of the ink was dropcasted
onto an area of 1 cm × 1 cm of graphite sheet and dried under air at 60 ◦ C for 1 h (catalyst loading
mg
∼ 0.24 cm2 ).

4.5. XRD Analysis

For the powder diffraction measurements, a Philips PW 1720 (Philips Analytical Technology
GmbH) with a copper anode (λ = 1.541 Å) and a 5 × 0.1 mm line focus was used. The sample was
placed in a pan made of brass. For data collection frames were measured for the duration of 10 s
per step in 0.02◦ intervals of 2θ among 5◦ and 91◦ . The data were collected with by using a graphite
secondary monochromator and a proportional counter.

4.6. TEM Lamella Preparation and TEM Analysis

For the TEM analysis 5 mg of powder (MoS2+δ /MoS2 ) were dissolved in 1 mL of THF and then a
30 min long ultra-sonic treatment applied. One drop of this solution was then put on a carbon TEM Grid.
The electron diffraction TEM investigations were performed with a Phillips CM12 (Philips Electron
Optics GmbH) at 120 kV. For the high resolution TEM analysis a FEI Titan (FEI, Hillsboro, Oregon,
USA) aberration corrected electron microscope with 300 keV was employed. For this microscope the
information limit in high vacuum is about 0.08 nm.

4.7. Accuracy of Lattice Spacing Measurement in TEM

The TEM was calibrated using a gold reference sample to get a reliable accuracy. The dependence
of the precision ∆d/d on defocus induced changes of diffuseness of the FFT reflections is much smaller
than the effect of a small diffraction vector length [42]. For each magnification, area of the FFT and
measured d value, the precision ∆d/d is calculated as a function of diffraction vector length and
presented in Table S5. For the average values in Table 2, the weighted average of each lattice spacing
dhkl is calculated from individual measurements, taking into account the error of the individual values.
The total error is the statistical deviation from the weighted average plus the highest systematic error
of the single measurements.
The TEM calibration was verified by using MoS2 nanoparticles. The resulting values for MoS2
are d(100) = (2.80 ± 0.06) Å and d(001) = (12.8 ± 0.7) Å. They are matching our XRD data and deviate
from the literature values of d(100) = 2.737 Å and d(001) = 12.294 Å obtained by x-ray diffraction in
Wildervanck et al. [30] by δ(100) = 2.3% and δ(001) = 4.2%. Within error, the determined MoS2 lattice
parameters are in agreement to these XRD results. Based on the calibration, the measurement of the
locally reduced d(100) value of MoS2+δ in this work is significant.

5. Summary
MoS2+δ was prepared by solvothermal synthesis, immobilized on electrodes and electrochemically
analyzed by cyclic voltammetry and chronopotentiometry. A comparison with MoS2 confirmed the
much higher activity of MoS2+δ . Hence the analyzed samples show the typical characteristics of
MoS2+δ which are reported in literature [4,11,14,18].
However, by means of HRTEM, XRD and electron diffraction, the structure of MoS2+δ is assigned
to a highly disordered variant of MoS2 with very small nanocrystal size and cluster like local defects.
The in-plane lattice parameter shows strong local variations and is reduced locally, whereas c-axis is
increased in average. Due to the nanocrystalline structure, a high concentration of edges with changed
Mo coordination and Mo–Mo distance is present. Furthermore, our HRTEM observations suggest that
Catalysts 2020, 10, 856 14 of 16

disorder within the MoS2 planes represent [Mo3 S13 ]2− cluster like defects. Such structural features are
also involved in the formation of disclinations. Altogether, this increases the ratio of coordinatively
modified Mo and is in accordance to the scaling of electrochemical activity with Mo. Our results imply
that S:Mo ratios > 2 are mainly important for the HER activity due to the processing induced nano-
and defect structure.

Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4344/10/8/856/s1,

Figure S1: Rietveld refinement of MoS2 , Figure S2: Electron diffraction pattern of MoS2.6 and MoS3.4 , Figure S3:
HRTEM images of MoS2 +δ used for the lattice parameter analysis in Table S5, Figure S4: HRTEM images of MoS2
and their FFTs, Figure S5: Reduced FFT of HRTEM image from Figure 5 A & B, Figure S6: SEM images of MoS2 and
MoS3.4 powder, Figure S7: Cyclic voltammetry of the two MoSx samples and MoS2 , Table S1: Comparison of dhkl
from XRD and SAD from Figures 2 and 3 for MoS2+δ with MoS2 , Table S2: Goodness parameters and correction
factor for the texture, Table S3: Refined lattice parameters and atom positions of MoS2 , Table S4: Overview of
the most intense lattice planes in electron diffraction of MoS2 , Table S5: Result from the FFT analysis of HRTEM
images of MoS2+δ .
Author Contributions: Conceptualization, E.R., S.H., P.K. and C.J.; investigation, E.R., M.-L.G., H.K., P.K., C.J.;
data curation, E.R., S.H., M.-L.G.; writing—original draft preparation, E.R., S.H., M.-L.G. and C.J.; writing—review
and editing, E.R., M.-L.G., P.K., C.J.; supervision, H.K., P.K. and C.J. All authors have read and agreed to the
published version of the manuscript.
Funding: This research was funded by Deutsche Forschungsgemeinschaft (DFG), within the national priority
program SPP 1613 “Fuels Produced Regeneratively Through Light-Driven Water Splitting”.
Acknowledgments: The authors would like to thank Max Baumung for his contribution to the TEM sample
preparation and Vladimir Roddatis for practical advices in TEM work.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Pellow, M.A.; Emmott, C.J.M.; Barnhart, C.J.; Benson, S.M. Hydrogen or batteries for grid storage? A net
energy analysis. Energy Environ. Sci. 2015, 8, 1938–1952. [CrossRef]
2. Chalk, S.G.; Miller, J.F. Key challenges and recent progress in batteries, fuel cells, and hydrogen storage for
clean energy systems. J. Power Sources 2006, 159, 73–80. [CrossRef]
3. Edwards, P.P.; Kuznetsov, V.L.; David, W.I.F.; Brandon, N.P. Hydrogen and fuel cells: Towards a sustainable
energy future. Energy Policy 2008, 36, 4356–4362. [CrossRef]
4. Morales-Guio, C.G.; Hu, X. Amorphous Molybdenum Sulfides as Hydrogen Evolution Catalysts. Acc. Chem. Res.
2014, 47, 2671–2681. [CrossRef]
5. Benck, J.D.; Chen, Z.; Kuritzky, L.Y.; Forman, A.J.; Jaramillo, T.F. Amorphous molybdenum sulfide catalysts
for electrochemical hydrogen production: Insights into the origin of their catalytic activity. ACS Catal. 2012,
2, 1916–1923. [CrossRef]
6. Liu, S.; Zhang, X.; Zhang, J.; Lei, Z.; Liang, X.; Chen, B. MoS2 with tunable surface structure directed by
thiophene adsorption toward HDS and HER. Sci. China Mater. 2016, 59, 1051–1061. [CrossRef]
7. Hinnemann, B.; Moses, P.G.; Bonde, J.; Jørgensen, K.P.; Nielsen, J.H.; Horch, S.; Chorkendorff, I.; Nørskov, J.K.
Biomimetic hydrogen evolution: MoS2 nanoparticles as catalyst for hydrogen evolution. J. Am. Chem. Soc.
2005, 127, 5308–5309. [CrossRef]
8. Jaramillo, T.F.; Jørgensen, K.P.; Bonde, J.; Nielsen, J.H.; Horch, S.; Chorkendorff, I. Identification of active
edge sites for electrochemical H2 evolution from MoS2 nanocatalysts. Science 2007, 317, 100–102. [CrossRef]
9. Seo, B.; Joo, S.H. Recent advances in unveiling active sites in molybdenum sulfide-based electrocatalysts for
the hydrogen evolution reaction. Nano Converg. 2017, 4, 19. [CrossRef]
10. Voiry, D.; Salehi, M.; Silva, R.; Fujita, T.; Chen, M.; Asefa, T.; Shenoy, V.B.; Eda, G.; Chhowalla, M. Conducting
MoS2 nanosheets as catalysts for hydrogen evolution reaction. Nano Lett. 2013, 13, 6222–6227. [CrossRef]
11. Ting, L.R.L.; Deng, Y.; Ma, L.; Zhang, Y.-J.; Peterson, A.A.; Yeo, B.S. Catalytic Activities of Sulfur Atoms in
Amorphous Molybdenum Sulfide for the Electrochemical Hydrogen Evolution Reaction. ACS Catal. 2016, 6,
861–867. [CrossRef]
12. Hsu, C.L.; Chang, Y.H.; Chen, T.Y.; Tseng, C.C.; Wei, K.H.; Li, L.J. Enhancing the electrocatalytic water
splitting efficiency for amorphous MoSx . Int. J. Hydrogen Energy 2014, 39, 4788–4793. [CrossRef]
Catalysts 2020, 10, 856 15 of 16

13. Kokko, M.; Bayerköhler, F.; Erben, J.; Zengerle, R.; Kurz, P.; Kerzenmacher, S. Molybdenum sulphides on
carbon supports as electrocatalysts for hydrogen evolution in acidic industrial wastewater. Appl. Energy
2017, 190, 1221–1233. [CrossRef]
14. Yan, Y.; Ge, X.; Liu, Z.; Wang, J.Y.; Lee, J.M.; Wang, X. Facile synthesis of low crystalline MoS2 nanosheet-coated
CNTs for enhanced hydrogen evolution reaction. Nanoscale 2013, 5, 7768–7771. [CrossRef]
15. Vrubel, H.; Merki, D.; Hu, X. Hydrogen evolution catalyzed by MoS3 and MoS2 particles. Energy Environ.
Sci. 2012, 5, 6136. [CrossRef]
16. Tran, P.D.; Tran, T.V.; Orio, M.; Torelli, S.; Truong, Q.D.; Nayuki, K.; Sasaki, Y.; Chiam, S.Y.; Yi, R.;
Honma, I.; et al. Coordination polymer structure and revisited hydrogen evolution catalytic mechanism for
amorphous molybdenum sulfide. Nat. Mater. 2016, 15, 640–646. [CrossRef]
17. Ye, Z.; Yang, J.; Li, B.; Shi, L.; Ji, H.; Song, L.; Xu, H. Amorphous Molybdenum Sulfide/Carbon Nanotubes
Hybrid Nanospheres Prepared by Ultrasonic Spray Pyrolysis for Electrocatalytic Hydrogen Evolution. Small
2017, 13, 1700111. [CrossRef]
18. Grutza, M.-L.; Rajagopal, A.; Streb, C.; Kurz, P. Hydrogen evolution catalysis by molybdenum sulfides
(MoSx ): Are thiomolybdate clusters like [Mo3 S13 ]2− suitable active site models? Sustain. Energy Fuels 2018, 2,
1893–1904. [CrossRef]
19. Weber, T.; Muijsers, J.C.; Niemantsverdriet, J.W. Structure of Amorphous MoS3 . J. Phys. Chem. 1995, 99,
9194–9200. [CrossRef]
20. Huntley, D.R.; Parham, T.G.; Merrill, R.P.; Sienko, M.J. An EXAFS study of the thermal decomposition of
molybdenum trisulfide. Inorg. Chem. 1983, 22, 4144–4146. [CrossRef]
21. Hibble, S.J.; Wood, G.B. Modeling the Structure of Amorphous MoS3 : A Neutron Diffraction and Reverse
Monte Carlo Study. J. Am. Chem. Soc. 2004, 126, 959–965. [CrossRef]
22. Mabayoje, O.; Wygant, B.R.; Wang, M.; Liu, Y.; Mullins, C.B. Sulfur-Rich MoS6 as an Electrocatalyst for the
Hydrogen Evolution Reaction. ACS Appl. Energy Mater. 2018, 1, 4453–4458. [CrossRef]
23. Müller, A.; Fedin, V.; Hegetschweiler, K.; Amrein, W. Characterization of amorphous substances by studying
isotopically labelled compounds with FAB-MS: Evidence for extrusion of triangular Mo3 IV clusters from a
mixture of 92 MoS3 and 100 MoS3 . J. Chem. Soc. Chem. Commun. 1992, 1795–1796. [CrossRef]
24. Hibble, S.J.; Feaviour, M.R.; Almond, M.J. Chemical excision from amorphous MoS3 ; a quantitative EXAFS
study. J. Chem. Soc. Dalton Trans. 2001, 0, 935–940. [CrossRef]
25. Li, Y.; Yu, Y.; Huang, Y.; Nielsen, R.A.; Goddard, W.A.; Li, Y.; Cao, L. Engineering the Composition and
Crystallinity of Molybdenum Sulfide for High-Performance Electrocatalytic Hydrogen Evolution. ACS Catal.
2015, 5, 448–455. [CrossRef]
26. Dave, M.; Rajagopal, A.; Damm-Ruttensperger, M.; Schwarz, B.; Nägele, F.; Daccache, L.; Fantauzzi, D.;
Jacob, T.; Streb, C. Understanding homogeneous hydrogen evolution reactivity and deactivation pathways
of molecular molybdenum sulfide catalysts. Sustain. Energy Fuels 2018, 2, 1020–1026. [CrossRef]
27. Luo, Z.; Ouyang, Y.; Zhang, H.; Xiao, M.; Ge, J.; Jiang, Z.; Wang, J.; Tang, D.; Cao, X.; Liu, C.; et al. Chemically
activating MoS2 via spontaneous atomic palladium interfacial doping towards efficient hydrogen evolution.
Nat. Commun. 2018, 9, 2120. [CrossRef]
28. Hu, J.; Huang, B.; Zhang, C.; Wang, Z.; An, Y.; Zhou, D.; Lin, H.; Leung, M.K.H.; Yang, S. Engineering stepped
edge surface structures of MoS2 sheet stacks to accelerate the hydrogen evolution reaction. Energy Environ.
Sci. 2017, 10, 593–603. [CrossRef]
29. Yin, Y.; Han, J.; Zhang, Y.; Zhang, X.; Xu, P.; Yuan, Q.; Samad, L.; Wang, X.; Wang, Y.; Zhang, Z.; et al.
Contributions of Phase, Sulfur Vacancies, and Edges to the Hydrogen Evolution Reaction Catalytic Activity
of Porous Molybdenum Disulfide Nanosheets. J. Am. Chem. Soc. 2016, 138, 7965–7972. [CrossRef]
30. Wildervanck, J.C.; Jellinek, F. Preparation and Crystallinity of Molybdenum and Tungsten Sulfides. Z. für
Anorg. und Allg. Chemie 1964, 328, 309–318. [CrossRef]
31. Colliex, C.; Cowley, J.M.; Dudarev, S.L.; Fink, M.; Gjønnes, J.; Hilderbrandt, R.; Howie, A.; Lynch, D.F.;
Peng, L.M.; Ren, G.; et al. Electron diffraction. In International Tables for Crystallography; International Union
of Crystallography: Chester, UK, 2006; pp. 259–429.
32. Michel, F.M.; Antao, S.M.; Chupas, P.J.; Lee, P.L.; Parise, J.B.; Schoonen, M.A.A. Short- To medium-range
atomic order and crystallite size of the initial FeS precipitate from pair distribution function analysis.
Chem. Mater. 2005, 17, 6246–6255. [CrossRef]
Catalysts 2020, 10, 856 16 of 16

33. Shyam, B.; Stone, K.H.; Bassiri, R.; Fejer, M.M.; Toney, M.F.; Mehta, A. Measurement and Modeling of Short
and Medium Range Order in Amorphous Ta2 O5 Thin Films. Sci. Rep. 2016, 6, 32170. [CrossRef] [PubMed]
34. Hall, B.D.; Zanchet, D.; Ugarte, D. Estimating nanoparticle size from diffraction measurements. J. Appl.
Crystallogr. 2000, 33, 1335–1341. [CrossRef]
35. Srolovitz, D.J.; Safran, S.A.; Homyonfer, M.; Tenne, R. Morphology of Nested Fullerenes. Phys. Rev. Lett.
1995, 74, 1779–1782. [CrossRef] [PubMed]
36. Nguyen, D.N.; Nguyen, L.N.; Nguyen, P.D.; Thu, T.V.; Nguyen, A.D.; Tran, P.D. Crystallization of Amorphous
Molybdenum Sulfide Induced by Electron or Laser Beam and Its Effect on H2 —Evolving Activities. J. Phys.
Chem. C 2016, 120, 28789–28794. [CrossRef]
37. Xi, F.; Bogdanoff, P.; Harbauer, K.; Plate, P.; Höhn, C.; Rappich, J.; Wang, B.; Han, X.; Van De Krol, R.; Fiechter, S.
Structural Transformation Identification of Sputtered Amorphous MoSx as an Efficient Hydrogen-Evolving
Catalyst during Electrochemical Activation. ACS Catal. 2019, 9, 2368–2380. [CrossRef]
38. Wu, L.; Longo, A.; Dzade, N.Y.; Sharma, A.; Hendrix, M.M.R.M.; Bol, A.A.; de Leeuw, N.H.; Hensen, E.J.M.;
Hofmann, J.P. The Origin of High Activity of Amorphous MoS2 in the Hydrogen Evolution Reaction.
ChemSusChem 2019, 12, 4383–4389. [CrossRef]
39. McDonald, J.W.; Friesen, G.D.; Rosenhein, L.D.; Newton, W.E. Syntheses and characterization of ammonium
and tetraalkylammonium thiomolybdates and thiotungstates. Inorg. Chim. Acta 1983, 72, 205–210. [CrossRef]
40. Li, Y.; Wang, H.; Xie, L.; Liang, Y.; Hong, G.; Dai, H. MoS2 nanoparticles grown on graphene: An advanced
catalyst for the hydrogen evolution reaction. J. Am. Chem. Soc. 2011, 133, 7296–7299. [CrossRef]
41. Cui, W.; Liu, Q.; Xing, Z.; Asiri, A.M.; Alamry, K.A.; Sun, X. MoP nanosheets supported on biomass-derived
carbon flake: One-step facile preparation and application as a novel high-active electrocatalyst toward
hydrogen evolution reaction. Appl. Catal. B Environ. 2015, 164, 144–150. [CrossRef]
42. Biskupek, J.; Kaiser, U. Practical considerations on the determination of the accuracy of the lattice parameters
measurements from digital recorded diffractograms. J. Electron Microsc. (Tokyo) 2004, 53, 601–610. [CrossRef]
[PubMed]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).