Calculation of MoS2 slabs morphology descriptors from transmission electron microscopy data revisited. Case study of the influence of citric acid and treatment conditions on the properties of MoS2/Al2O3
Graphical abstract
Revisited method of calculation of MoS2 slabs length and stacking in the supported catalysts has been used in a systematic study of the influence of citric acid and treatment conditions on the sulfide morphology and HDS activity.
Introduction
Molybdenum sulfide–based hydrotreating catalysts are among the most widely used catalytic materials for hydrotreating of oil feedstock. These catalysts consist of alumina-supported Co(Ni)-promoted MoS2 slabs with the active sites located on the slabs edges [1].
The information on the structure of active phase is difficult to obtain due to its small size and complexity. While the 2D shape of MoS2 slabs perpendicular to [002] axis is impossible to assess with conventional transmission electron microscopy (TEM), the slabs length and stacking are easily available from TEM and often presented as main morphology descriptors. In many studies statistical analysis of length and stacking numbers of MoS2 slabs had been reported, sampling usually between 200 and 500 individual sulfide particles. The mean values of slab length (L) and stacking layer number (S) are calculated from the frequency histograms as the corresponding first moments for individual distributions; S and L being treated as independent variables [2], [3], [4]. Variations of < S > and < L > are evoked to explain observed catalytic activity as a function of preparation route, treatment conditions and the nature of support, additive and/or promoter [5], [6], [7], [8]. The same particle-wise first moment treatment is applied to the tungsten sulfide – based catalysts [9], [10] Even without calculating the mean values, in most cases one-dimensional histograms on their own right give a correct qualitative morphology representation in the series of similar catalysts (as in the works cited above).
However when the first moments <L> and <S> are applied to calculate some specific values, they cease to be satisfactory descriptors. That is particularly important when L and S are strongly correlated or in some subtle cases when the effects are not very strong and the conclusions depend on the fine details of MoS2 morphology. Then we need to treat L and S as dependent variables. One of such subtle cases concerns the influence of citric acid on the properties of non-promoted sulfide catalysts.
Citric acid (CA) is widely used as additive to prepare Co(Ni)-Mo sulfide catalysts. However the possible mechanisms of CA action are still questionable. Beneficial effect of CA seems well established for the Co(Ni)-Mo(W) promoted catalysts supported on alumina [11], [12], [13] or mesoporous silica [14], [15]. There, CA is supposed to stabilize bimetallic species, to affect the sulfidation sequence and to favor the formation of promoted CoMoS phase and/or to stabilize shorter MoS2 slabs. Fewer works dealt with the non-promoted MoS2/Al2O3. The benefit of CA in such systems seems less clear as it depends on the preparation conditions. Okamoto et al. evidenced a positive effect of CA, if introduced after air calcination of a high-loaded impregnated oxide [16]. Recently, Castillo-Villalón et al. [17] and Chen et al. [18], [19] studied non-promoted MoS2/Al2O3 catalysts in the presence of CA. The IR spectroscopy of adsorbed CO (IR-CO) showed a strong increase of 2070 cm−1 band, attributed to CO adsorbed on S-edges of MoS2 slabs and therefore it has been concluded that 2D shape of MoS2 slabs is affected [17], [18], [19]. According to Ref. [18], beside the 2D shape CA slightly affects the MoS2 particles length and stacking. However, Ramirez et al. in [17] found little or no effect of CA on the TEM- observed morphology and on the activity of the non-promoted catalysts in HDS of dibenzothiophene.
Our previous studies suggest a strong impact of the treatment conditions on the slabs length, stacking, 2D morphology and sulfur coverage of MoS2 slabs edges in both bulk [20] and supported [21] catalysts. We inquired how the interplay between the CA addition and the variations of treatment conditions would influence the TEM-observed MoS2 morphology and other morphology-dependent properties. The initial goal was to find the conditions maximizing the impact of CA. Therefore, the properties of MoS2/Al2O3 catalysts prepared with or without citric acid and activated under variable conditions are systematically studied in this work, and intrinsic activity values are compared using revisited approach to the calculation of <L> and <S> from the TEM data.
Section snippets
Preparation of the catalysts
Gamma alumina with specific surface area 259 m2/g was applied as support. Prior to the preparations, the alumina powder (80–120 μm fraction) was calcined in dry air flow at 400 °C for 2 h. The impregnated samples were kept for 48 h under ambient conditions then oven dried for 12 h at 80°, 125° and 150 °C respectively, in order to study the effect of drying temperature.
In the main series twenty four catalysts have been prepared, forming a full factor experiment for the following parameters: two Mo
Revisited method of MoS2 slabs mean stacking and length calculation
Frequency histograms of slabs length (L) and stacking (S) are widely used to represent MoS2 catalysts. In the literature, the first momenta of L and S distributions are calculated using the Eq. (1):Where <M> is an estimated quantity (<L>, or <S>), Ni in the number (frequency) of particles or slabs having the length within the discrete interval i (nm) or integer stacking i, N− total number of particles or slabs. For <S>, according to the Eq. (1) any isolated MoS2 particle (stack) is
Conclusions
In this work we revisit the method of evaluation of MoS2 slabs morphology from the data of transmission electron microscopy and propose to treat both length and stacking using surface weighted distributions, instead of particle-wise first momentum. Moreover, as shows analysis of conditional distributions, the slabs length L and stacking S are strongly correlated, that formally prohibits their statistical analysis as independent variables. Corrected values of <L> and <S> are significantly larger
Acknowledgements
The author thanks Ruben Checa (IRCELYON) who performed carbon-sulfur analyses and Laetitia Oliviero (ENSICaen, France) who suggested me to study the influence of the drying conditions.
References (47)
- et al.
J. Catal.
(2001) - et al.
J. Catal.
(2016) - et al.
Appl. Catal. A Gen.
(1993) - et al.
J. Catal.
(2001) - et al.
J. Catal.
(2002) - et al.
J. Catal.
(2016) - et al.
Appl. Catal. A Gen.
(2011) - et al.
J. Molec. Catal. A Chem.
(2015) - et al.
J. Mol. Catal. A Chem.
(2008) - et al.
Appl. Catal. B Env.
(2013)