Sulfide catalysis without coordinatively unsaturated sites: Hydrogenation, cis-trans isomerization, and H2/D2 scrambling over MoS2 and WS2

Article abstract of DOI:10.1021/ja3074903

Simple test reactions as ethene hydrogenation, 2-butene cis-trans isomerization and H₂/D₂ scrambling were shown to be catalyzed by MoS₂ and WS₂ in surface states which did not chemisorb oxygen and were, according to XPS analysis, saturated by sulfide species. This is a clear experimental disproof of classical concepts that require coordinative unsaturation for catalytic reactions to occur on such surfaces. It supports new concepts developed on model catalysts and by theoretical calculations so far, which have been in need of confirmation from real catalysis.

Full text of DOI:10.1021/ja3074903

Communication
pubs.acs.org/JACS
Sulfide Catalysis without Coordinatively Unsaturated Sites:
Hydrogenation, Cis−Trans Isomerization, and H /D Scrambling over
2
2
MoS and WS
Thomas Drescher, Felix Niefind, Wolfgang Bensch, and Wolfgang Gru
2
2
#
§
§
,#
̈
nert*
#
Laboratory of Industrial Chemistry, Ruhr University Bochum, P.O. Box 102148, 44780 Bochum, Germany
Institute of Inorganic Chemistry, Christian Albrechts University Kiel, Otto-Hahn-Platz 10, 24118 Kiel, Germany
§
*
S Supporting Information
capacity (OCS) which might indicate coordinative unsatura-
tion. XPS intensity data and reactivity experiments confirm that
the catalytic effect originates from a saturated surface
containing hydrogen. Although it is not yet possible to identify
the sites operating in our catalysts with any of the theoretical
models, our results strongly support the new concepts
according to which saturated sites make major contributions
to sulfide catalysis.
ABSTRACT: Simple test reactions as ethene hydro-
genation, 2-butene cis−trans isomerization and H /D
2
2
scrambling were shown to be catalyzed by MoS and
2
WS in surface states which did not chemisorb oxygen and
2
were, according to XPS analysis, saturated by sulfide
species. This is a clear experimental disproof of classical
concepts that require coordinative unsaturation for
catalytic reactions to occur on such surfaces. It supports
new concepts developed on model catalysts and by
theoretical calculations so far, which have been in need
of confirmation from real catalysis.
MS2+x were prepared by thermal decomposition of
15
(
NH ) MS in inert gas. Inspired by Afanasiev’s work, series
4 2 4
of samples with varying S/M ratios were made by varying the
16,17
decomposition temperatures.
Here, data will be presented
from samples with S/Mo = 2.56 and S/W = 2.24 (“MoS2.55”,
“
WS2.25”), while more details and data on the remaining
16,17
he catalytic properties of MoS and WS (“MS ”) have
materials will appear elsewhere.
included elemental analysis, XRD, N2 physisorption and
TEM. XRD revealed the 2H-MS structure of the sulfides
(Figure S1), their small particle sizes in stacking direction (on
average 3.4 and 3.8 nm for MoS2.55 and WS2.25), and eventually
lattice plane bending (slight shift of (002) reflections to lower
angles). These observations are confirmed by TEM images
(Figure S2).
Basic characterization
2
2
2
T
attracted tremendous attention due to their importance
1
for hydrorefining catalysis. There has been general agreement
2
that these catalysts operate via coordinatively unsaturated sites
2
3
(
cus) at the edges of their layered structures, and detailed
assignments as to which reaction requires which extent of
4
−6
coordinative unsaturation were proposed.
In the past
decade, different views on the active sites on MS2 have
emerged from work with model catalysts and from theoretical
OCS and catalysis were studied in a setup allowing easy
switch between flow and recycle modes. Samples were reduced
6
studies. On single, fully saturated MoS slabs, a metallic state
2
extending on the basal plane adjacent to the slab edge was
in dilute H (10% in He) or thermoevacuated at various
2
7
,8
observed by Scanning Tunneling Microscopy (STM), and
theoretical calculations suggested potential catalytic activity of
temperatures (xyz K: R , V ) before the OCS capacity was
xyz
xyz
measured (pulse mode, 273 K). The adsorbed oxygen was
removed by a mild hydrogenation at 473 K (IR473), which did
not change the OCS capacity. Thus, the Vxyz pretreatment
8
these “brim” sites. Although this was supported by STM
images showing adsorbed thiophene to be cleaved on them in
9
,10
presence of atomic hydrogen,
the gap between model
involved a final contact with H at 473 K. Catalytic studies were
2
situation and real catalysis remains formidable, calling for
experimental support for the relevance of these structures
under real conditions. At the same time, theoretical studies have
demonstrated the ability of various sulfur species at saturated
made in recycle mode (see Supporting Information). The
absence of any oxygen traces that could interfere with the OCS
measurements was scrupulously examined (cf. refs 16, 17).
Thus, a MoS2.15 sample that yielded zero OCS capacity after
thermoevacuation (V723 without subsequent IR473) was
recovered from the reactor via a glovebox, and studied by
XPS after sample transfer without air contact: No indication for
oxidation neither of Mo nor of S were found, and the minor O
MoS2 edges to adsorb and dissociate H without any
2
10−12
coordinative unsaturation,
although the adsorption of
hydrocarbons has not been discussed so far. Another recent
concept at variance with the classical cus models is based on the
observation that the surface layers absorbs carbon during
interaction with real feeds, and the activity is assigned to Mo
1
s sample present already in the as-prepared material was not
increased. This measurement showed the near-surface region
stronger enriched in sulfur than even in the initial sample ((S/
Mo)XPS = 2.52 vs 2.42, bulk ratios −1.91 and 2.17,
1
3,14
carbide like active species.
We are reporting here that both MoS and WS can catalyze
2
2
simple test reactions like ethene hydrogenation (EH), cis−trans
isomerization of 2-butene (CT), and H /D scrambling (H D )
Received: July 30, 2012
2
2
2
2
in surface states that do not exhibit any oxygen chemisorption
Published: November 5, 2012
©
2012 American Chemical Society
18896
dx.doi.org/10.1021/ja3074903 | J. Am. Chem. Soc. 2012, 134, 18896−18899

Journal of the American Chemical Society
Communication
Figure 1. Catalytic activity and OCS capacity of MoS2.55 after different pretreatments. (a) Ethene hydrogenation, Tcat = 463 K; (b) cis−trans
isomerization of 2-butene, Tcat = 400 K; (c) H /D scrambling, T = 463 K. Activation energies resulting from different pretreatments shown in the
2
2
cat
panels.
Figure 2. Catalytic activity and OCS capacities of WS2.25 after different pretreatments. (a) Ethene hydrogenation, Tcat = 463 K; (b) cis−trans
isomerization of 2-butene, Tcat = 400 K; (c) H /D scrambling, T = 463 K. Activation energies resulting from different pretreatments shown in the
2
2
cat
panels.
respectively): Apparently, thermoevacuation leads to a
saturated surface, the vacancies being distributed into the bulk.
During the activation treatments, textural properties (BET
surface area, stacking height) and composition of the materials
varied in a complex manner, which has no obvious relation to
the catalytic properties. Upon reduction, the BET surface area
In Figure 1, activity data (first-order rate constants at a
reference temperature Tcat) of MoS2.55 are summarized and
compared with the OCS capacities. The range of activation
energies E observed is also reported. After a pronounced
A
maximum at Tred = 523 K (56.6 μmol/g), the OCS capacity fell
to zero at 823 K. After V723, the OCS capacity was zero as well,
directly after evacuation, after the IR473 step (see above) and
after catalysis.
2
of MoS2.55 exhibited a monotonous increase from 5 m /g
2
2
(
R473) to >60 m /g (R ), while a shallow maximum at 72 m /
773
The hydrogenation activity (Figure 1a) initially increased
g was found for WS2.25 after R573 (see Figure S3). Stacking
heights started to grow moderately at reduction temperatures
with T , in contrast to the drastically decreasing OCS capacity
red
6
as reported in Polyakov et al. Upon more severe reduction, the
Tred > 573 K (MoS2.55) and >823 K (WS ). Reduction caused
2.25
1
6,17
hydrogenation activity dropped as well, but remained
measurable even at Tred = 873 K where the OCS capacity
had vanished. With the same zero OCS capacity, MoS2.55
sulfur loss, but with different patterns.
already after R573, the S/Mo ratio of MoS2.55 went through a
minimum of 1.65 after R773 returning to 1.87 after R923
Falling to 1.91
,
exhibited an almost 20-fold activity when activated by V723
.
whereas the S/W ratio of WS2.25 never fell below 2.
Analogously, a monotonous decay above S/M = 2 was
observed with MoS2+x of small S excess, and a pronounced
All these treatments resulted in activation energies scattering
around 37 kJ/mol (±3 kJ/mol) although the OCS capacity
span a range from ≈57 to 0 μmol/g. The CT activity (Figure
1b) declined with increasing reduction severity as described in
1
6,17
S/W minimum with WS2+x of large S excess.
We assign this
6
to sulfide mobility at higher temperatures (>723 K) where
Polyakov et al. A dramatic change of the activation energy
1
6,17
more stable structures may delay loss of sulfur.
This
from ≈60 to ≈30 kJ/Mol was noted after R873. After V723, the
activity was much larger, but the activation energy remained
close to that after R873. In H D (Figure 1c), activity started
mobility may also explain the pronounced peaks in the
temperature dependence of the OCS capacities (Figures 1
and 2). Thermoevacuation resulted in nearly stoichiometric
samples (S/Mo − 1.98, S/W − 1.96).
2
2
high and at a very low activation energy (15 kJ/mol) after R
5
73
6
as found in our earlier work. Already at T = 673 K, the
red
1
8897
dx.doi.org/10.1021/ja3074903 | J. Am. Chem. Soc. 2012, 134, 18896−18899

Journal of the American Chemical Society
Communication
activity had decreased and the activation energy had increased
four times. After V723, the activity was even smaller, but EA
activation energy. We rather suggest that hydrogenation activity
may not be related to oxygen chemisorption at all. We propose
therefore that EH always proceeds on saturated sites: those
observed in surface states without OCS capacitybrim sites or
hydrogen activating sulfide species. As to CT, the different
activation energies in catalyst states with high or zero OCS
capacity (Figure 1b) suggest that the reaction rate is
remained the same as after R673
.
With other initial S/Mo values, similar tendencies were
observed, with characteristic deviations though in details
(Figure S4, see also refs 16, 17). CT activity was always higher
after V723 than after any reductive activation, and H D activity
2
2
1
was lower. Hydrogenation activity after V723 exceeded that after
contributed from different sites, one of which might be a M
6
activation in H at low initial S/Mo ratio and dropped below
site as proposed earlier. The case of H D reaction is very
2
2
2
the latter at high initial S/Mo ratio.
complex, and further work is in progress. There is competition
between different reaction mechanisms as well, the contribu-
tions of which may be very sensitive to details of the activation
procedure, which results in unexpected switches of E . H D
2
can also proceed in absence of OCS capacity, but at a relatively
low reaction rate (Figures 1c and 2c).
Analogous data for WS2.25 are depicted in Figure 2. The OCS
on tungsten sulfides exhibited similar trends as on MoS2+x, but
peak capacities were lower and required higher reduction
A
2
1
6,17
temperatures.
WS2.25 had a peak OCS capacity of 19 μmol/
g at Tred = 823 K, a temperature where MoS2.55 did not adsorb
oxygen any more (Figures 1 and 2). The OCS capacity of
WS2.25 did not vanish at Tred ≤ 1023 K although its decrease
was obvious. After V873, the OCS capacity was zero as in case of
the MoS2+x catalysts.
While the contribution of saturated sites to catalysis over
MS appears to be well supported by our data (including those
2
from three more MoS and two more WS catalysts (refs 16,
2
2
17, Figure S4)), it is not yet possible to discriminate
contributions from them and from other types of sites because
there is not yet a tool to count those saturated sites that are
involved in the catalytic reactions. As mentioned with regard to
EH, we are skeptical on the relevance of OCS as an indicator
for active sites. Theoretical work has cast doubt on the
Unlike with MoS2.55, the hydrogenation activity of WS2.25
nearly paralleled the OCS capacity with variation of Tred
(Figure 2a). Again, however, a sizable activity was measured
with the thermoevacuated sample, which did not adsorb
oxygen. In the limits of experimental error, all treatments
resulted in the same activation energy irrespective of the OCS
4
,18
traditional mechanisms of H D
by showing the minor
and of multiple vacancies on
Mo, and the variety of sulfur species capable of splitting
2
2
1
9,20
capacity achieved. Changes in E occurred for the CT and
stability of Mo−H species
A
20
H D reactions (Figure 2b,c) as observed with MoS (Figure
2
2
2.55
10−12
1
b,c). In CT, the activation energy dropped to ca. 50% at
higher reduction temperatures and remained low after V873
Figure 2b). As with MoS2.55, the activity after thermoevacua-
dihydrogen
may well give rise to the observed changes in
the activation energy without any need for cus. The above-
1
(
mentioned reference to M sites which would host the half-
tion was the highest observed for CT. The behavior of the
H D reaction was very complex (Figure 2c). A shift from lower
hydrogenated intermediate of CT is an unproved suggestion.
Still, one may wonder why EH should not proceed on cus on
surfaces offering huge OCS capacities (Figure 1a). We found
evidence that OCS may count not only cus, but cover also a
secondary reaction with adsorbed H. Actually, using our
2
2
to higher activation energy (≈45 to ≈75 kJ/mol) occurred with
increasing reduction severity, but E reproducibly shifted back
A
to the lower level in a narrow range of Tred around 923 K. As
2
‑
with MoS2.55, the H D activity after thermoevacuation was
textural data and a stoichiometry of 1 O per missing S , one
may estimate that with our peak OCS capacity of 57 μmol/g
(thought to arise from edge planes) the edge Mo atoms are all
2
2
smaller than achieved after reductive activations.
These data show that EH, CT, and H D can proceed on
2
2
1
6,17
MoS and WS surfaces which do not adsorb oxygen. Oxygen
exposed.
the IR473 treatment in D and repeated the OCS run, we could
This is certainly unrealistic. When we performed
2
2
chemisorption is generally accepted to indicate coordinative
unsaturation of Mo(W) sites, and our XPS analysis clearly
supports that surfaces with zero oxgen chemisorption capacity
are saturated. Therefore, our reactions obviously can proceed
on saturated surfaces, that is, on exposed sulfide species. This is
what the recent studies under model conditions and theoretical
investigations predict, although it is not yet possible to identify
which kind of saturated sites is operative in our real catalysts. It
includes actually also Mo carbide sites although their formation
from the adsorbed hydrocarbon species should be less likely
under our mild reaction temperatures, which are well below
those in hydrorefining processes.
The invariability of the EH activation energy over a wide
range of activation conditions is a further remarkable result.
This “universal” activation energy changes with the initial S/Mo
ratio, from 73 kJ/mol for MoS2.15 to ca. 40 kJ/mol for MoS2.55
and MoS2.75 (refs 16, 17, Figure S4). The range of OCS
capacities achieved by these activations is sometimes enormous
as seen in Figure 1a where the activation energy does not
respond to a change of the OCS capacity between 0 and almost
2
17
indeed detect D O in the effluent, see Drescher et al. This
2
evidence (though as yet qualitative) for a rapid reaction of
chemisorbed oxygen with surface hydrogen at room temper-
ature suggests that OCS overestimates the cus to an extent
depending on the availability of surface hydrogen. The details
are subject to ongoing research. Because of the minor role
ascribed by us to OCS as a tool for the detection of active sites,
this finding does not invalidate our major conclusions.
Another important observation concerns the role of surface
hydrogen in hydrogenation. The appreciable (in case of MoS
2
.15
17
actually very high ) hydrogenation activity after thermoevac-
uation contradicts observations in Polyakov et al. where low
6
hydrogenation activities after V723 were obtained in catalytic
runs immediately following the activation, that is, without
intermediate contact with hydrogen by IR473 (see above).
Reproducing this, we found an extremely low activity of
MoS2.15 directly after thermoevacuation, which increased by
almost 2 orders of magnitude when the catalyst was treated in
H at 473 K (for details see ref 17). In this treatment, the
2
6
0 μmol/g. Similar observations were made with the WS
catalyst adsorbed hydrogen without acquiring any OCS
capacity. This observation suggests the existence of a surface
hydrogen species acting as a co-catalyst in hydrogenation, quite
reminiscent to the report of the Besenbacher group about the
2+x
1
6
samples. Such result can be reconciled with a competition
between two different reaction paths (over 3-fold unsaturated
and over saturated sites) only if both routes present the same
1
8898
dx.doi.org/10.1021/ja3074903 | J. Am. Chem. Soc. 2012, 134, 18896−18899

Journal of the American Chemical Society
Communication
importance of surface hydrogen for the adsorption of thiophene
(7) Lauritsen, J. V.; Helveg, S.; Lægsgaard, E.; Stensgaard, I.; Clausen,
B. S.; Topsøe, H.; Besenbacher, F. J. Catal. 2001, 197, 1.
9
,10
on the single-slab model catalysts. In our case, however, the
coexistence of ethene with hydrogen in the gas phase appeared
(
8) Bollinger, M. V.; Lauritsen, J. V.; Jacobsen, K. W.; Norskov, J. K.;
Helveg, S.; Besenbacher, F. Phys. Rev. Lett. 2001, 87, 196803.
9) Lauritsen, J. V.; Nyberg, M.; Vang, R. T.; Bollinger, M. V.;
17
to prevent the formation of the co-catalyst. The experiment
also shows that OCS does count coordinative unsaturation:
The surface hydrogen did not react with the oxygen offered in
the OCS run. Obviously, there is no reaction between surface
hydrogen and oxygen if the latter is not activated by adsorption
onto cus.
(
Clausen, B. S.; Topsøe, H.; Jacobsen, K. W.; Lægsgaard, E.; Nørskov, J.
K.; Besenbacher, F. Nanotechnology 2003, 14, 385.
(10) Lauritsen, J. V.; Nyberg, M.; Nørskov, J. K.; Clausen, B. S.;
Topsøe, H.; Lægsgaard, E.; Besenbacher, F. J. Catal. 2004, 224, 94.
(11) Dinter, N.; Rusanen, M.; Raybaud, P.; Kasztelan, S.; da Silva, P.;
Toulhoat, H. J. Catal. 2010, 275, 117.
In conclusion, ethene hydrogenation, 2-butene cis−trans
isomerization, and H /D scrambling were observed over MoS
2
(12) Prodhomme, P.-Y.; Raybaud, P.; Toulhoat, H. J. Catal. 2011,
2
2
2
(
3
(
80, 178.
13) Kelty, S. P.; Berhault, G.; Chianelli, R. R. Appl. Catal., A 2007,
22, 9.
14) Chianelli, R. R.; Berhault, G.; Torres, B. Catal. Today 2009, 147,
75.
and WS2 catalysts in surface states in which oxygen
chemisorption, a proven indicator for coordinative unsatura-
tion, did not occur. Obviously, the reactions can proceed on
saturated sites, which is of considerable importance for the
understanding of catalysis on sulfides although the surface
states of our catalysts certainly deviate from those under real
hydrorefining conditions due to differences in temperature and
conditions of the gas phase. Ethene hydrogenation exhibits the
same activation energy over catalyst states with widely varying
2
(15) Afanasiev, P. J. Catal. 2010, 269, 269.
(16) Drescher, T. Ph.D. Thesis, Ruhr University Bochum, 2011.
̈
(17) Drescher, T.; Niefind, F.; Bensch, W.; Kienle, L.; Grunert, W. J.
Catal. 2012, submitted for publication.
(18) Polyakov, M.; van den Berg, M. W. E.; Hanft, T.; Poisot, M.;
Bensch, W.; Muhler, M.; Grunert, W. J. Catal. 2008, 256, 126−136.
19) Byskov, L. S.; Bollinger, M.; Norskov, J. K.; Clausen, B. S.;
Topsoe, H. J. Mol. Catal. A 2000, 163, 117.
20) Byskov, L. S.; Nørskov, J. K.; Clausen, B. S.; Topsøe, H. J. Catal.
999, 187, 109.
OCS capacities, whereas changes in E were found for cis−
̈
A
(
trans isomerization and H /D scrambling, indicating the
2
2
relevance of more than one type of active sites. These
observations suggest that ethene hydrogenation may occur on
saturated sites only, whereas other types of sites are also
available for the remaining reactions. Ethene hydrogenation is
strongly accelerated by the presence of a specific surface
hydrogen species, the identity of which is being further
investigated.
(
1
ASSOCIATED CONTENT
■
*
S
Supporting Information
Sample preparation, characterization techniques, description of
catalytic experiments. Characterization data of MoS2.55, WS2.25
:
XRD in initial state, TEM after R573 (MoS2.55), R873 (WS2.25),
development of BET surface area with reduction temperature.
Summary of catalytic data of ethene hydrogenation for all MoSx
catalysts. This material is available free of charge via the
Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
w.gruenert@techem.rub.de
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Financial support by the German Science Foundation (grants
No. Gr 1447/15, BE 1653/11) is gratefully acknowledged. We
thank Prof. L. Kienle (Kiel) for providing TEM images of our
catalysts.
REFERENCES
■
(
1) Topsoe, H.; Clausen, B. S.; Massoth, F. E. Hydrotreating Catalysis
Science and Technology; Springer Verlag: New York, 1996; Vol. 11.
(
(
(
(
2) Siegel, S. J. Catal. 1973, 30, 139.
3) Tanaka, K. I.; Okuhara, T. J. Catal. 1982, 78, 155.
4) Tanaka, K.-I.; Okuhara, T. Catal. Rev.−Sci. Eng. 1977, 15, 249.
5) Jalowiecki, L.; Aboulaz, A.; Kasztelan, S.; Grimblot, J.; Bonelle, J.
P. J. Catal. 1989, 120, 108.
6) Polyakov, M.; Poisot, M.; Bensch, W.; Muhler, M.; Gru
Catal. 2008, 256, 137.
(
̈
nert, W. J.
1
8899
dx.doi.org/10.1021/ja3074903 | J. Am. Chem. Soc. 2012, 134, 18896−18899