Catalysis Letters Vol. 82, No.

1±2, September 2002 (#2002) 13

Hydrothermal syntheses and catalytic properties of dispersed

molybdenum sul®des
Elodie Devers, Pavel Afanasiev *, Bernadette Jouguet, and Michel Vrinat
Institut de Recherches sur la Catalyse, 2 Avenue A. Einstein, 69626 Villeurbanne, CeÂdex, France

Received 31 January 2002; accepted 28 March 2002

Reactions of ammonium thiomolybdate under mild hydrothermal conditions were studied, and the range of conditions leading to the
dispersed MoS2 product has been determined. The morphology and the catalytic properties of the hydrothermal MoS2 preparations have
been compared with those of reference MoS2 produced from the thermal decomposition of thiosalts. It has been shown that the length of
the slabs and their stacking number of the crystallites of hydrothermal MoS2 preparations di€er considerably from those in the reference
sul®des. The morphological di€erences strongly in¯uence the hydrogenation to hydrodesulfurization activity ratio in these systems.
KEY WORDS: hydrothermal; MoS2 ; morphological.

1. Introduction and the temperature ranged from 473 to 573 K. Relatively

low temperatures and short reaction time were applied in
Molybdenum disul®de (MoS2 ) is of considerable order to obtain high-surface-area materials.
interest in several ®elds such as catalysis, photochemistry The solids were characterized by X-ray di€raction
or production of lubricating oils. In recent years, a (XRD), elemental chemical analysis, scanning electron
number of works appeared on the preparation of un- microscopy (SEM), and surface-area measurements.
supported dispersed molybdenum sul®de catalysts, High-resolution transmission electron microscopy
using thermal decomposition of molecular precursors (HREM) combined with EDX analysis were done on a
[1±3]. Earlier we prepared MoS2 dispersions in aqueous JEOL 2010 microscope with an accelerating voltage of
solutions [4,5] or in hydrothermal and solvothermal 200 keV.
brines [6]. The use of hydrothermal conditions provides Thiophene hydrodesulfurization (HDS) and tetraline
an original synthetic possibility for elaboration of new hydrogenation (HYD) were chosen as model reactions
compounds and for preparation of known substances for comparison of the catalytic properties. Thiophene
with particular morphology. HDS was carried out in vapor phase in a ®xed-bed
The present work is focused on the study of the microreactor operated in the dynamic mode at 100 kPa
products of reactions of (NH4 )2 MoS4 (ATM) precursor of hydrogen (thiophene pressure 2.4 kPa, total ¯ow
under mild hydrothermal conditions. Our goal was to 6 l=h). The catalyst charge was of 0.1 g. For the tetralin
study the products of reaction in reaction mixtures simi- gas-phase HYD reaction, the range of temperatures
lar to that used previously in [6]. Hydrothermal prepara- studied was 523±573 K, the hydrogen pressure was
tion of MoS2 with controlled morphology, if accessed by 4.5 MPa, the tetralin vapor partial pressure was 8.9 kPa
this method, might present considerable interest for and the H2 S pressure was 84 kPa. In all catalytic tests,
catalysis, both in applied and fundamental aspects. the speci®c rate was calculated at 573 K in the steady
state after 16 h of stream according to the equation
R ˆ FX=m, where R is the speci®c rate (mol gÿ1 sÿ1 ), F
2. Experimental is the molar ¯ow rate of the reactant (mol sÿ1 ), X is the
conversion of reactant, and m is the catalyst weight (g).
Hydrothermal preparations were carried out in a stain- In both catalytic tests the products were analyzed by
less steel bomb of 0.25 l volume. To this vessel were added gas chromatography.
solutions of 2.8 g (NH4 )2 MoS4 (ATM) in 100 ml of water.
Reaction was carried out in a nitrogen atmosphere, or
under 20 bar of H2 S. In some experiments, 50 ml of 3. Results and discussion
(NH4 )2 S solution or 12 g of elementary sulfur was
added. The reaction duration varied from 15 min to 4 h 3.1. Products of hydrothermal syntheses

* To whom correspondence should be addressed. It was found previously [6] that under hydrothermal
E-mail [email protected] conditions ATM might be transformed to solids

1011-372X/02/0900-0013 # 2002 Plenum Publishing Corporation

14 E. Devers et al. / Hydrothermal syntheses and catalytic properties of dispersed molybdenum sul®des

Table 1
Conditions of hydrothermal syntheses and the products obtained

Ref. T (K) Time (h) Sul®dizing agent Phases XRD

493-1-S 493 1 H2 S (NH4 )2 Mo3 S13 , (MoSx )

473-4-S 473 4 H2 S (NH4 )2 Mo3 S13 , (MoSx )
533-4-S 533 4 H2 S MoS2
573-4-S 573 4 H2 S MoS2
513-F 513 0.2 ± MoS2
513-F-N 513 0.2 (NH4 )2 S (NH4 )2 Mo3 S13
513-1-N 513 1 (NH4 )2 S (NH4 )2 Mo3 S13 , MoS2
513-1-N5 a 513 1 (NH4 )2 S (NH4 )2 Mo3 S13 , (NH4 )2 Mo2 S12
513-1-S8 513 1 S8 MoS2

a
The amount of (NH4 )2 S divided by 5.

possessing original properties. In the present work, vary- much longer MoS2 sheets (table 2). The hydrothermally
ing the reaction conditions and the source of sulfur, we prepared solids have speci®c surface areas between 80
followed the composition of the solid products. Di€erent and 100 m2 /g and are thermally stable, being able to
solids were obtained, including non-stoichiometric withstand heating in nitrogen to 773 K without consider-
sul®des MoSx with x varying from 3 to 6, condensed able loss of the speci®c surface area. We believe that
thiomolybdates (NH4 )2 Mo2 S12 and (NH4 )2 Mo3 S13 or hydrothermal conditions promote the planar growth of
highly dispersed stoichiometric MoS2 having anisotropic the MoS2 sheets. As a result, transmission microscopy
morphology and speci®c surface areas of 80±120 m2 /g. shows mostly low-energy (002) planes exposed in the
The most important results are shown in table 1. The hydrothermal samples, explaining therefore the resis-
samples are referenced as XXX-Y-A, where X is the tivity of their texture to the thermal treatments. Note
reaction temperature, Y is the time of reaction (F that the nature of the textural stabilization in this case
means fast, about 0.2 h) and A is the sul®dizing agent. is di€erent from that reported earlier [1±5], which was
In some cases the products are mixtures of di€erent due to the persistence of some carbonaceous impurities.
Mo-containing phases; in other cases pure compounds Beside MoS2 , the products often formed in our
were obtained. Over 533 K all the preparations yield as synthesis are ammonium thiomolybdates. Very pure
a main product molybdenum sul®de, MoS2 . Note that (NH4 )2 Mo3 S13 compound can be obtained rapidly,
formation of MoS2 under hydrothermal conditions using ammonium sul®de as a sulfur source, while the
occurs at lower temperatures than thermal decomposi- Mo precursor might be either ATM or ammonium hep-
tion of the same ATM precursor at atmospheric pressure tamolybdate. The needle-like and often hollow crystals
under nitrogen or hydrogen ¯ow. Indeed, thermal of (NH4 )2 Mo3 S13 are observed in SEM (®gure 2).
decomposition of thiomolybdates occurs only in the Their heating under nitrogen or hydrogen ¯ow at
temperature region 573±693 K [7±9]. Moreover, recently
the hydrothermal syntheses of MoS2 from the oxide
precursors using various sul®dizing agents have been
reported at temperatures even lower than 533 K [10,11].
It seems that MoS2 is the most stable sul®de compound
at these conditions; therefore it is produced whatever
the precursors used and the source of sulfur. Using the
ATM precursor in which molybdenum is already
bound to sulfur, MoS2 might be prepared hydro-
thermally at 513 K without the addition of other
sulfur-containing compounds (513-F sample). In this
case formation of MoS2 occurs due to decomposition
of ATM analogously to that occurring upon thermal
treatment, but due to hydrothermal conditions the
decomposition temperature is lowered and the morphol-
ogy of the sul®de strongly changed.
Hydrothermal MoS2 presents in SEM the agglomer-
ates of anisotropic particles of rather uniform shape
and size as depicted in ®gure 1. Compared with the sam-
ples prepared from the thermal decomposition of ATM, Figure 1. SEM micrograph (17 000) of the hydrothermally-prepared
the hydrothermal samples have higher surface area and MoS2 , 513-FSS sample.
 E. Devers et al. / Hydrothermal syntheses and catalytic properties of dispersed molybdenum sul®des 15

Table 2
Properties of the unsupported sul®de catalysts before and after catalytic tests

Catalyst Initial After HDS After HYD Stacking Fringe

S (m2 /g) length
S (m2 /g) S/Mo S (m2 /g) S/Mo XRD TEM (nm)

513-F 86 80 1.7 65 1.9 6 7 47  5

513-1-S 95 92 1.75 71 1.9 ± ± ±
Ex-th2 54 16 1.8 19 1.88 ± ± ±
Ex-th1 67 22 1.8 36 1.95 4 5 51

S is speci®c surface area. S/Mo is atomic ratio determined by elemental analysis.

673 K produces MoS2 having the same needle-like mor- this con®guration, each sulfur ion is bound to three
phology. In the narrow range of conditions, ammonium Mo ions, which results in a weak interaction with the
thiodimolybdate (NH4 )2 Mo2 S12 was formed, in the form next layer. Though it is generally accepted that the cata-
of plate-like crystals. To obtain this compound, the lytically active CUS centers are localized at the edges of
amount of (NH4 )2 S applied should be decreased. layers [12,13], the role of stacking (number of MoS2
The acid±base properties of the reaction mixture seem layers in crystallites) for the catalytic properties is not
to play a determining role. The mixtures containing yet completely clear. In the patches of MoS2 fringes,
(NH4 )2 S have basic reaction, pH being between 9 and two kinds of topologically and energetically unequal
10, depending on the amount of ammonium sul®de CUS molybdenum atoms can be distinguished according
added. Varying this last amount, pure trimolybdate, of to their position in the stacks: those located inside the
its mixture with dimolybdate was obtained. By contrast, stacks (called ``edge sites''), and those at the outer
in the cases when H2 S, S8 or no sulfur compound was layers (``rim sites''). According to the ``rim±edge''
added, the mixture after the reaction is weakly acidic model proposed by Daage and Chianelli [14], rim and
(pH 3). In general, basic pH stabilizes (NH4 )2 Mo3 S13 edge sites have considerably di€erent hydrogenation
compound, whereas softly acidic medium is more favor- and hydrodesulfurization properties. For the reactions
able for the formation of microcrystalline MoS2 and of dibenzothiophene, they concluded that HYD
amorphous MoSx sul®des. occurred exclusively on the rim sites, whereas HDS was
obtained on both the rim and the edge sites. Therefore,
3.2. Catalytic properties of the solids the e€ects of structural sensitivity might be observed in
these reactions for materials with varying fringe lengths
Insight into the catalytic properties of MoS2 requires and stacking numbers. Originally, the rim±edge model
consideration of its anisotropic structure. Within one was based on the properties of the solids obtained from
layer, its structure can be viewed as a two-dimensional the thermal decomposition of MoS3 at di€erent tem-
macromolecule with each Mo ion surrounded by six peratures. Using a hydrothermal preparation technique,
sulfur anions in a trigonal prismatic arrangement. In we provide an alternative way of preparing MoS2 with
varying morphology. Our initial goal was to compare
HYD and HDS properties of such systems in terms of
the rim±edge model. Such a study of di€erently prepared
unsupported catalysts was undertaken by Iwata et al.
[15,16] for the reactions of dibenzothiophene HDS and
methylnaphthalene HYD. They suggested that in addi-
tion to rim and edge sites there are also active in¯ection
points of basal planes. In our work, the properties of
hydrothermal MoS2 are compared with those of the
solids obtained from the thermal decomposition of
thiosalts. We compare the activities of unsupported
sul®des in the gas-phase HDS and HYD reactions of
small molecules, which are thiophene and tetralin
respectively.
For the catalytic study we used two hydrothermal
samples: 513-1-S8, prepared from ATM and elementary
sulfur, and 513-F, synthesized from sole ATM by its
hydrothermal decomposition. The properties of these
Figure 2. SEM micrograph (220) of the hydrothermally-prepared two samples are similar: both consist of long stacked
(NH4 )2 Mo3 S13 , 513-1H sample. patches of well-stacked MoS2 fringes as in ®gure 3. The
16 E. Devers et al. / Hydrothermal syntheses and catalytic properties of dispersed molybdenum sul®des

Figure 3. HREM micrograph (400 000) of the 513-FSS sample after the Figure 4. HREM micrograph (400 000) of the Ex-th2 sample after the tet-
tetralin HYD catalytic test. ralin HYD catalytic test.

composition determined prior to catalytic tests, after important share of the rim sites. The analysis of breadth
activation in the H2 S/H2 ¯ow at 673, was MoS2 ‡ x , of the (002) XRD line and estimation of stacking from
x ˆ 0:04±0.05. In table 2, the morphology of hydrother- the HREM images give the same qualitative result,
mal specimens and the reference sul®des is described, as though the absolute values of average stacking numbers
measured by transmission microscopy, analysis of somewhat di€er (table 2).
powder X-ray di€raction lines and nitrogen adsorption. Compared with the ex-thiomolybdates, the speci®c
The total speci®c surface area depends on the number activity of hydrothermal MoS2 is higher in the thiophene
of slabs in the MoS2 crystallites (stacking) and the lateral HDS but lower in the tetralin HYD, as presented in
extent of MoS2 layers (fringe length). table 3. The intrinsic activity of hydrothermal samples
The two reference samples for comparison were is always lower, because of their high speci®c surface
prepared from the decomposition of thiosalts. The ®rst, area, in which a great contribution of catalytically
Ex-th1, is obtained from the decomposition of ATM at inactive basal planes is present. However, the share of
673 K in hydrogen for 4 h. The second, Ex-th2, was pre- border (rim or edge) sites for the hydrothermal samples
pared from the (NH4 )2 Mo2 S12 using the same treatment is about 10 times lower than for ex-thiomolybdates.
conditions as for Ex-th1. The samples have similar initial Therefore, the intrinsic activity should be lower for the
surface areas of 67 and 55 m2 /g respectively, and have hydrothermal sul®des by an order of magnitude, which
the composition MoS2 ‡ x , x ˆ 0:06±0.07. Both ex- is not the case. The reason for such a discrepancy
thiomolybdate solids are constituted by shorter fringes might be in the di€erent accessibility of catalytic centers
with less stacking than the hydrothermal ones (®gure for the reacting molecules or in the non-zero activity of
4). Though the correct statistical treatment of HREM curved basal planes, as suggested by Iwata et al. [15,16].
images is dicult for the unsupported samples because As follows from the results of the catalytic study
of stacks overlapping, it seems obvious that the edge presented in table 3, qualitative di€erences of catalytic
sites are relatively more abundant for the hydrothermal activity exist between hydrothermal and thermal
samples, whereas the ex-thiosalt solids expose a more decomposition-derived sul®des which are of the same

Table 3
Reactivity of sul®des in thiophene HDS and tetralin HYD at 573 K

Catalyst Tetralin HYD Thiophene HDS

Vs (10ÿ8 mol/g s) Vi (10ÿ8 mol/m2 s) C0 /C2ÿ ratio Vs (10ÿ8 mol/g s) Vi (10ÿ8 mol/m2 s)

513-FSS 12.2 0.17 0.1 44.6 0.48

513-1S 9.1 0.14 0.09 18.0 0.23
Ex TS 1 28.0 0.78 0.27 13.5 0.61
Ex TS 2 15.7 0.82 0.47 9.0 0.60

Vs is the speci®c reaction rate. Vi is the intrinsic reaction rate. C0 /C2ÿ is the butane to butene ratio.
 E. Devers et al. / Hydrothermal syntheses and catalytic properties of dispersed molybdenum sul®des 17

Figure 5. Thiophene HDS reaction network.

kind, for all four preparations studied. The variation in reports. However, their stability is high and, being free
the HYD/HDS activity ratio is remarkable: for the of carbon, they are also free from its in¯uence on the
hydrothermally-prepared samples the HDS activity catalytic properties, whatever the latter. Such a set of
(either speci®c or intrinsic) is always greater than the solids gives a good opportunity to study the structural
HYD activity, whereas for the solids obtained by ther- sensitivity of the HYD and HDS reactions. We see that
mal decomposition the ratio of activities is inverse. In the HYD/HDS activity ratio was inverted in these
agreement with the HYD/HDS activity ratios observed experiments as a function of the catalyst morphology.
in two di€erent reactions, the selectivity in the thiophene Therefore, control of Mo sul®de catalyst selectivity via
HDS is also very di€erent at comparable conversions. As the change of its morphology is indeed possible.
follows from its less hydrogenating properties, both
hydrothermal samples have at least three times higher
butene to butane ratio (table 3). Therefore, without References
resolving the kinetic equations we may state that the
direct HDS pathway without intermediate formation of [1] N. Rueda, R. Bacaud and M. Vrinat, J. Catal. 169 (1997) 404.
tetrahydrothiophene (®gure 5) prevails in these speci- [2] G. Alonso, M. Del Valle, J. Cruz, A. Licea-Claverie, V. Petranovskii
and S. Fuentes, Catal. Lett. 52 (1998) 55.
mens over the consecutive hydrogenation±HDS route. [3] Y. Yoneyama and C. Song. Catal. Today 50(1) (1999) 19.
It seems that this di€erence depends only on the sul®de [4] I. Bezverkhyy, P. Afanasiev and M. Lacroix, Inorg. Chem. 39 (2000)
morphology, because the initial and measured S/Mo 5416.
ratios after catalytic tests are very similar for two types [5] P. Afanasiev, C. Geantet, C. Thomazeau and B. Jouget, Chem.
of preparation. As for eventual impurities, their in¯uence Commun. 12 (2000) 1001.
[6] P. Afanasiev, G.-F. Xia, G. Berhault, B. Jouguet and M. Lacroix,
is unlikely since all the samples were prepared from the Chem. Mater. 11 (1999) 3216.
decomposition of the same ATM precursor. [7] A. Leist, S. Stauf, S. LoÈken, E.W. Finckh, S. LuÈdtke, K.K. Unger, W.
Assenmacher, W. Mader and W. Tremel, J. Mater. Chem. 8(1) (1998)
241.
4. Conclusions [8] S.J. Hibble and M.R. Feaviour, J. Mater. Chem. 11(10) (2001)
2607.
[9] V.P. Fedin, J. Czyzniewska, J. Prins and Th. Weber, Appl. Catal. 213
Hydrothermal synthesis allows one to obtain molyb- (2001) 123.
denum sul®de compounds with original morphology. [10] J. Bokhimi, A. Toledo, J. Navarrete, X.C. Sun and M. Portilla, Int.
The reaction temperature and the acidity of the reaction J. Hydr. Energy 26(12) (2001) 1271.
mixture are key parameters, determining the nature of [11] Y. Peng, Z. Meng, C. Zhong, J. Lu, W. Yu, Z. Yang and Y. Qian,
J. Solid State Chem. 159 (2001) 170.
the products obtained. Varying these parameters, pure [12] R.R. Chianelli, Int. Rev. Phys. Chem. 2 (1982) 127.
phases of needle-like (NH4 )2 Mo3 S13 of highly dispersed [13] S. Kasztelan, H. Toulhoat, J. Grimblot and J.P. Bonnelle, Appl. Catal.
MoS2 with very long layers have been prepared. They 13 (1984) 127.
are interesting model solids with which to study the [14] M. Daage and R.R. Chianelli, J. Catal. 149(2) (1994) 414.
nature of catalytic function in the sul®de systems. The [15] Y. Iwata, K. Sato, T. Yoneda, Y. Miki, Y. Sugimoto, A. Nishijima and
H. Shimada, Catal. Today 45 (1998) 353.
surface areas of hydrothermal MoS2 solids obtained in [16] Y. Iwata, Y. Araki, K. Honna, Y. Miki, K. Sato and H. Shimada,
this work (about 80±100 m2 /g) are somewhat lower Catal. Today 65 (2001) 335.
than those of carbon-stabilized samples from previous