J. Chem. SOC., Faraday Trans. I, 1986, 82, 2423-2434 Correlation between Hydrodesulphurization Activity and Reducibility of Unsupported MoS,-based Catalysts promoted by Group VIII Metals Sandor GobolosJ Qin WuJ Olivier Andre', Francis Delannay and Bernard Delmon* UniuersitP Catholique de Louuain, Groupe de Physico-Chimie MinPrale et de Catalyse, Place Croix du Sud I , 1348 Louuain-la-Neuve, Belgium Unsupported FeMo, CoMo and NiMo sulphide hydrodesulphurization (HDS) catalysts have been prepared by one or several different methods: homogeneous sulphide precipitation (HSP), inverse HSP (IHSP), comacer- ation (CM) and coprecipitation (CP). They have been characterised by (i) differential thermal analysis (DTA) during temperature-programmed reduc- tion/sulphidation (TPR/S) and subsequent air oxidation (TPO), (ii) temperature-programmed sulphur extraction (TPSE), (iii) XPS and (iv) XRD.For catalysts prepared by the HSP, IHSP and CP methods the onset temperature of TPR/S decreases, and the threshold temperature of oxidation, the amount of released H,S in TPSE and the HDS activity all increase in the sequence Mo < FeMo < CoMo < NiMo. XPS reveals a change in the Co 2p,,, and Ni 2p,,, binding energies for promoted catalysts with respect to pure Cogs, and NiS, respectively. The results demonstrate a strong correlation between catalytic activities in HDS of thiophene and hydrogen- ation of cyclohexene on the one hand, and the reducibility of the catalysts on the other. The industrial importance of hydrosulphurization (HDS) and hydrotreating and the vigorous expansion of these processes are encouraging numerous attempts to correlate the HDS activity of both supported and unsupported catalysts with their physicochemical properties.Characterisation of catalysts by Mossbauer spectroscopy,' 0, chemisorption,2 NO chemisorption combined with i.r. ~pectroscopy,~ e.s.r. ~pectroscopy,~ temperature- programmed reduction, temperature-programmed hydrogen de~orption,~ conductivity measurements6 and magnetic susceptibility measurements' all provide interesting corre- lations with the catalytic activity. The catalytic activity was also reported to be related to the number of sulphydryl groups8 and amount of labile sulphur.5* 9+11 The experimental material available for this type of study is extremely rich, in the sense that samples differing by many variables have been prepared and studied. These variables are (i) the methods of preparation; (ii) presence or absence of carrier and its nature; (iii) composition, defined by r, the atomic ratio of the Group VIII metal to the Group VIII plus Group VI metals; and (iv) the nature of the Group VIII (Co, Ni or Fe) and Group VI (Mo, W) metals.It is surprising that relatively little use has been made of the possibilities offered by series of catalysts containing different Group VIII metal promoters. The similarity in characteristics and catalytic activities of the corresponding catalysts is considerable. A criterion for a good correlation between HDS (or hydrogenation, HYD) activity and Budapest, Pusztaszeri ut 59-61, Hungary.P.B. 2653, People's Republic of China. 7 On leave from the Central Research Institute for Chemistry of the Hungarian Academy of Sciences, 1025 $ On leave from the Beijing Municipal Chemical Industry Research Institute, Cheng Fu, Haidian, Beijing, 24232424 Activity of MoS,-based Catalysts physicochemical properties is that it should hold for all six systems comprising one Group VI metal with one Group VIII metal. The present study is an attempt in this direction. A comparison is made between unsupported FeMo, CoMo and NiMo HDS catalysts. We shall study the correlations between (i) the reduction/sulphidation of catalyst precursors under an H,S-H, mixture, (ii) the oxidation of the catalysts under air, (iii) the reduction (sulphur extraction) of the catalysts under H, and (iv) the HDS activity.XPS and XRD results will also be included. The role of sulphur in unsupported HDS catalysts will be discussed. Experimental Catalyst Preparation Precursors The chemicals used for the preparations were Merck products (pro analysi). The various stages of preparation (precipitation, evaporation and drying) were carried out under an argon atmosphere at 343 K. The unsupported sulphide catalyst precursors were prepared by four different methods. All methods involve the mixing of chosen proportions of compounds of molybdenum and Group VIII metal (M) in a solution of (NH,),S. The S/(M +4Mo) atomic ratio in the mixture was > 1.5 times that needed for the complete sulphidation of M and Mo into MS and MoSi-. These methods were as follows : (a) the homogeneous sulphide precipitation (HSP) method:12 a mixed solution of the Group VIII metal nitrate and ammonium heptamolybdate (AHM) was added to a solution of (NH4),S; (b) the inverse HSP (IHSP) method:', a solution of (NH,),S was added to a mixed solution of the Group VIII metal nitrate and AHM; ( c ) the comaceration (CM) method:'* powdered MOO, and the Group VIII metal oxide (e.g.Co,O,) were allowed to react with a solution of (NH,),S; (d) the coprecipitation (CP) mefhod:l5 a mixture of MS and MoS, (or mixed sulphides) was coprecipitated by adding a solution of the Group VIII metal nitrate into a previously prepared solution of ammonium thiomolybdate (ATM). Details of the experimental procedures have been described elsewhere. l5 Unsupported CoMo sulphide precursors with atomic ratios r = Co/(Co+Mo) = 0.1 and 0.3 were prepared by the HSP, IHSP and CM methods. Unsupported NiMo sulphide catalyst precursors with r = 0.1 and 0.3 were prepared by the HSP method.Unsupported FeMo sulphide catalyst precursors with r = 0.15 and 0.3 were prepared by the CP method using iron(II1) nitrate. In order to guarantee the precipitation of iron sulphide by the sulphur of ATM and thus to provide an intimate contact between iron and molybdenum, only a 3% excess of sulphur was used in this case, i.e. S/(Fe+4Mo) = 1.03 instead of 1.50. A precursor of pure molybdenum disulphide (r = 0) was also prepared by the HSP method. Reduct ion/ Sulphida t ion Treatment The active catalysts were prepared from the above precursors by a reduction/sulphidation treatment.The precursors were heated to 673 K at a rate of 0.33 K s-l under a 1.1 cm3 s-l flow of 15% H,S-H, at atmospheric pressure. After 4 h at 673 K, the samples were cooled to room temperature under a flow of argon. The catalysts will be designated in terms of the metal(s) present, followed by the method of preparation and the composition ratio r (e.g. Mo-HSP-0.0, FeMo-CP-0.15, CoMo-CM-0.3). Catalytic Activity Measurement The activity in HDS of thiophene and hydrogenation (HYD) of cyclohexene was measured on 0.5 g of catalyst (grain size 0.1-0.2 mm) under 3 MPa, at 573 K using aS . Gobolos, Q. Wu, 0. Andrk, I;. Delannay and B. Delmon 2425 feed mixture containing 69.5 wt % cyclohexane, 30.0 wt % cyclohexene and 0.5% thiophene.The mass flow rate of the liquid feed was 48 g h-l. As exposure to air could not be avoided during the loading of the reactor, samples were treated in situ for 1 h under 3 MPa, 4% H2S-H, at 573 K prior to activity tests. Samples of the product were analysed every half-hour for 8 h. Steady-state activities were always obtained after 3-4 h. The intrinsic activities are expressed as follows : mol mP2 s-l wc xc HYD=- Mc mSa where W, and W, are the mass flow rates (g s-l), MT and M , are the molecular weights, XT and X c are the conversions (at steady state after 8 h on stream) for thiophene and cyclohexene, respectively, Sa (m2 g-l) is the surface area of the used catalyst and m is the weight of the catalyst. Other details on activity measurements are described elsewhere fi Characterization Surface Area Measurement B.E.T. surface areas of fresh and used catalysts (after 8 h on stream) were determined by gravimetric measurement of nitrogen adsorption.Temperature-programmed Reactions Differential thermal analysis (DTA) of temperature-programmed reduction/sulphidation (TPR/S) of the dry precursors under a 15% H,S-H, mixture, and of the subsequent temperature-programmed oxidation (TPO) of the sulphided catalysts under air were performed under conditions identical to those described elsewhere.” Temperature-programmed sulphur extraction (TPSE) experiments were carried out at atmospheric pressure in a flow reactor.l8 The catalyst samples (0.2 g) were first resulphided in situ under 15% H,S-H, at 673 K for 2 h. After purging with argon at the same temperature for 1 h and cooling to room temperature, they were heated at a rate of 0.17 K s-l to 773 K in a flow of 0.5 cm3 s-l of pure H,.The H2S produced was measured by a thermal-conductivity detector. XPS After the activity test, samples were collected and stored in iso-octane, and pressed, also under a film of iso-octane, into the cupules used for XPS measurements. This procedure enabled to protect the samples from contact with air. The cupules covered with a meniscus of the solvent were then introduced into the vacuum chamber of the spectrometer where the solvent was evacuated. The effectiveness of this procedure in protecting sulphide catalysts from oxidation has already been proved. l6 The binding energies (Eb) were referred to the contaminant carbon peak (C 1s = 285.0 eV).Differences in Eb between levels Mo 3d5/, and S 2p [AE,(Mo-S)] and M 2p,/, and S 2p [AEb(M-S); M = Fe, Co or Nil were also calculated. S 2p/Mo 3d and M 2p/Mo 3d intensity ratios were converted into atomic ratios S,/Mo (t = total) and M/Mo using sensitivity factors proposed in the literature.ls2426 Activity of MoS,-based Catalysts Table 1. B.E.T. surface area, XRD and catalytic activity results intrinsic activityb / lo-* mol m-* s-l surface area/m2 g-l -~ ha catalyst fresh used /nm HDS HYD HYD/HDS M 0-HSP-0.0 FeMo-CP-0.15 COMO-CM-0. 1 COMO-IHSP-0. 1 COMO-HSP-0. 1 NiMo-HSP-0.1 FeMo-CP-0.3 COMO-CM-0.3 COMO-IHSP-0.3 COMO-HSP-0.3 NiMo-HSP-0.3 102.4 59.2 45.8 22.5 35.8 24.1 40.4 26.0 13.7 20.2 - 26.7 35.9 12.8 3.5 8.3 7.7 27.4 6.9 14.4 9.3 16.6 5.8 7.7 - - 7.6 6.6 5.5 5.2 5.5 0.82 1.78 1.96 5.39 4.63 2.94 2.34 4.71 8.41 7.62 13.6 19.2 24.1 23.8 37.7 29.3 52.5 31.5 37.7 76.8 57.4 59 1 23.4 13.5 12.1 7.0 6.5 17.9 13.5 8.1 9.1 7.5 43.6 a Average crystallite sizes estimated by XRD.Steady state activities after 8 h on stream. XRD X-Ray diffractograms were recorded using Cu Ka radiation and a monochromator to suppress the fluorescence from cobalt. The average crystal sizes were evaluated from the broadening of the (002) (h) lines for MoS,-containing catalysts using a k value of 1.077 in the Scherrer equation and assuming that the crystallites were free from strain. Results Catalytic Activities and B.E.T. Surface Areas B.E.T. surface areas of fresh and used catalysts and intrinsic activities are given in table 1.Surface areas always decrease during catalytic testing, especially in the case of catalysts with r = 0.1. HDS and HYD activities increase by the addition of Fe, Co or Ni. This increase in activity is always higher for catalysts with r = 0.3 than with r = 0.1 or 0.15. The promoting effect of various Group VIII metals for both HDS and HYD increases as follows : FeMo < CoMo < NiMo. The only exception is catalyst NiMo-HSP-0.1, which exhibits a lower HDS activity than catalyst CoMo-HSP-0.1. The HDS activity of catalysts CoMo-HSP-0.1 and CoMo-HSP-0.3 is always higher, by a factor of 2, than that of catalysts CoMo-CM-0.1 and CoMo-CM-0.3. The HYD/HDS ratio (selectivity) increases in the following order: CoMo < FeMo < MoS, < NiMo (the only exception is again catalyst NiMo-HSP-0.1).Physicochemical Characterization Temperat we-programmed React ions Fig. 1 presents the DTA curves for TPR/S and TPO of samples Mo-HSP-0.0 and NiMo-HSP-0.3. The onset temperatures of reduction/sulphidation ( TR,s) and oxidation (To) (T, and &, respectively, as in previous papers)l7? l9 are very much affected by the presence of the Group VIII metal. For more details the reader is referred to our previous papers. Table 2 summarizes the values of TR,s and To for all samples.S. Gobolos, Q. Wu, 0. Andre!, I;. Delannay and B. Delmon 2427 'R / S I I I I I I 3 00 500 700 TI K Fig. 1. DTA curves of temperature-programmed reduction/sulphidation of precursors and subsequent oxidation of Mo-HSP-0.0 and NiMo-HSP-0.3 catalysts.Table 2. Temperature-programmed reductiona and XPSb results ~ ~ ~ ~ ~ ~ TE,s To TPSE area Eb(s 2p) AEb(M-S)" catalyst /K /K (arb. units) St/Moc M / M O ~ . ~ /eV /ev -~ ~~~ ~ ~~ Mo-HSP-0.0 FeMo-CP-0.15 COMO-CM-0. 1 COMO-IHSP-0. 1 COMO-HSP-0. 1 NiMo-HSP-0.1 FeMo-CP-0.3 COMO-CM-0.3 COMO-IHSP-0. 3 COMO-HSP-0.3 NiMo-HSP-0.3 59 1 543 577 515 518 528 533 57 1 51 1 503 505 600 618 598 625 618 642 608 603 625 630 698 1.5 - - 5.0 5.0 8.0 15.0 - 1.93 1.96 - 1.91 1.95 2.17 1.89 2.1 1 2.04 2.09 - 0.22 0.08 0.13 0.10 0.52 0.16 0.37 0.43 0.48 162.8 162.9 162.9 162.5 162.4 162.9 162.6 162.5 162.2 162.3 - 545.9 616.6 616.8 692.4 545.9 61 6.9 617.0 617.1 692.6 a From ref. (17) (19). The precision of such quantitative XPS ratio measurements is & 5 % . The M/Mo atomic ratios calculated from the chemical compositions are 0.11, 0.18 and 0.43 for catalysts with Y = 0.1, 0.15 and 0.3, respectively." AE,(M-s) for CO,S,,~~ NiS20 and Fe,-, S21 are 616.1, 691 .O and 545.8 eV, respectively. Measurements on used catalysts. The threshold temperature T,,, of reduction/sulphidation of molybdenum-containing catalyst precursors is shifted towards lower temperatures in the presence of promoters. This effect is especially important for CoMo and NiMo catalysts with Y = 0.3, and for catalysts prepared by the HSP and IHSP methods. In contrast, less influence on the onset2428 Activity of MoS,-based Cutalysts 1 1 I I 1 I 400 600 800 TIK Fig. 2. Temperature-programmed sulphur extraction curves of catalysts referred to unit surface areas.temperature of reduction/sulphidation is observed in the case of Fe addition and of the CM method of preparation. The onset temperature of oxidation, To, increases as follows : Mo-HSP-0.0 z CoMo- CM-r < FeMo-CP-r < CoMo-HSP(1HSP)-r < NiMo-HSP-r. A change in the concen- tration of promoter markedly affects oxidation only in the case of Ni-containing catalysts. The results of temperature-programmed sulphur extraction measurements are shown in fig. 2. The amount of released H2S increases in the following order: Mo-HSP- 0.0 < CoMo-CM-0.3 z FeMo-CP-0.3 < CoMo-HSP-0.3 z NiMo-HSP-0.3. The main characteristics of the spectra are twofold : a small peak appears between 450 and 550 K for catalysts Mo-HSP-0.0, CoMo-CM-0.3 and FeMo-CP-0.3, and a large one appears with a maximum around 700 K for catalysts CoMo-HSP-0.3 and NiMo-HSP-0.3.The specific TPSE spectral areas are also listed in table 2. The S,/Mo and M/Mo atomic ratios, the binding energy of the S 2p level and the AEb(M-s) binding-energy differences are also reported in table 2. The S,/Mo ratio is slightly higher for catalysts with r = 0.3 than with r = 0.1. The M/Mo ratios are close to the bulk values for catalysts prepared by the HSP(1HSP) and CP methods. The measured Co/Mo ratio is significantly lower than the bulk ratio for catalyst The AE,,(Mo-S) values of 66.9-67.0 eV (not listed in table 2) observed for all catalysts were equal to that of pure M0S2.16 AEb(Ni-s) and AEb(co-s) values are significantly higher (by 1.4-1.6 and 0.5-1 .O eV) than those of pure NiSZ0 and Co,S,,16 respectively.No change in AEb(Fe-s) between FeMo catalysts and Fe,-, SZ1 was observed. COMO-CM-0.3.S. Gobolos, Q. Wu, 0. Andre!, F. Delunnay and B. Delmon 2429 XRD Average crystallite sizes of MoS, in the catalysts calculated from XRD results are also listed in table 1. Crystallite sizes of MoS, seem to be significantly higher in catalysts CoMo-CM-0.3 and FeMo-CP than in other samples. Discussion Textural Properties The crystallite sizes listed in table 1 correspond to higher surface areas than actually measured by B.E.T. techniques. This is due to the fact that MoS, presents a turbostratic structure22 where the particle size is larger than the domain size, with the consequence that a large proportion of the ‘domain’ boundaries is not accessible to the N, molecules in the B.E.T.measurement. The average crystallite size of MoS, decreases only slightly in the presence of Co or Ni in catalysts prepared by the HSP and IHSP methods. Candia et al.12 have found a more significant decrease in the particle size of MoS, by the addition of Co in unsupported catalysts prepared by the HSP method, We feel that this contradiction could be due to a difference in the details of the preparation of the samples. Both cobalt, in CMI4 and CP19 samples, and iron, in CP samples,1g increase the crystallite size of MoS,. This suggests that Group VIII metal sulphides decrease the disorder normally observed in the MoS, structure. Delannay et al.23 and Thakur et al.24 also reported that the addition of a small amount of Group VIII metal (r < 0.02) brings about a marked increase in crystallinity of MoS, in unsupported catalysts prepared by the CM method.They assumed that a small amount of promoter facilitates the growth of MoS, ~rystallites.~~* 24 However, the present results suggest that the same phenomena can occur at even higher promoter concentrations (r = 0.3) in the CMl49 2 3 9 24 and CPl5* l9 catalysts. However, the influence of promoter dispersion on the crystallinity of MoS, still needs further elucidation. Surface-area losses during catalytic tests are probably due to both crystal growth, favoured by a highly reducing atmo~phere,,~ and some blocking of the micropores by coke deposition.26 The higher the initial surface area and molybdenum content, the greater the loss.16 Catalytic Activity It is worthwhile to note the (low) promotion of both HDS and HYD activity of MoS, by iron in catalysts prepared by the CP method. This is in agreement with the results of Thakur et ~ l ., , ~ who have first reported a synergy between Fe and Mo in unsupported HDS catalysts prepared by the CM method. As already reported, the HDS activity of catalysts CoMo-HSP-r is higher than that of catalysts CoMo-CM-r.12 There is no significant difference in the activity of catalysts prepared by the HSP and IHSP methods, probably owing to the similarities of their structure and texture. The promoting effect of various Group VIII metals for both HDS and HYD (Fe < Co < Ni) changes in a similar way for HSP or CP and CM24 catalysts. The behaviour of catalyst NiMo-HSP-0.1 is not very surprising, since no or only slight promotion of HDS and HYD at low promoter concentrations ( r z 0.1) has been reported for both supported21 and unsupportedll? 24 NiMo sulphide catalysts.The HYD/HDS ratios confirmed again that CoMo sulphide or NiMo sulphide catalysts are the best choices when, respectively, HDS or HYD activity should be favoured.2430 Activity of MoS,-based Catalysts Physicochemical Characterization Before discussing in detail the results of physicochemical characterizations, it is worth recalling our present knowledge about the location of promoter atoms in unsupported MoS,-based HDS catalysts prepared by the different methods used in this study. The catalysts prepared by the CM method are essentially biphasic, containing separate MoS, and the thermodynamically stable MS, (e.g.Cogs8, Ni,S,, Felpz S) There is much similarityz8> 29 between the HSPl2 and IHSP13 samples. Topsrae et ~ 1 . ~ 9 30 have shown using Mossbauer spectroscopy and XRD that, in unsupported CoMo sulphide catalysts prepared by the HSP method,12 Co is present in two forms: Cogs, and a MoS,-like phase which was termed CO-MO-S.~*~O Very recently, infrared spectroscopy3~ 31 and analytical electron 32 have proved that, in the Co-Mo-S structure, Co atoms are located on the edges of MoS, slabs, most likely in substitutional or interstitial positions. Therefore, Co atoms cover, at least partly, the Mo sites on the edges of MoS, cry~tallites~~. 32 and modify their catalytic33 and physicochemical properties,l* 34 including, presumably, the metal-sulphur bonding.35 The existence of F~-Mo-S~~ and N~--Mo--S~~ structures in HDS catalysts was also established.In addition to FeS and Fel-,S, Fe-Mo-S was also detected by Mossbauer spectroscopy in unsupported catalysts prepared by the CP method.31 phases.14, 15, 23, 24, 27 Tempera t we-programmed React ions A detailed discussion of the TPR/S and TPO results is presented in ref. (17) and (19). Note that the exothermic effects, observed in the TPR/S experiments between 500 and 700 K, are due to the transformation of MoS, and/or Mo oxysulphides into MoS,, accompanied by the release of elemental sulphur and the crystallisation of MoS,.~~ The highly dispersed M-S species catalyse the reduction/sulphidation of Mo-O,S, species in catalyst precursors prepared by the HSP, IHSP and CP methods.Co and Ni sulphides are particularly efficient in promoting the reduction of Mo-containing species.lg In the TPO experiments a threshold temperature of 600 K is observed for the oxidation of pure MoS, under these ~0nditions.l~ Since the oxidation of MoS, proceeds from the edges,39 the lower reactivity of MoS, towards oxygen in the presence of Ni and Co is probably due to the covering of the edges of MoS, crystallites by promoter 31. 32 The origin of this reduced affinity towards oxygen is difficult to ascertain as long as the mechanism of this oxidation remains to be elucidated. One may speculate that the onset of oxidation corresponds to oxidation of the topmost atom exposed on the edge. In the case of pure MoS,, this atom is sulphur bound to molybdenum, which oxidizes into SO,.In the case of promoted catalysts, the first exposed sulphur atom may desorb before being oxidised into SO, (owing to a lower bond strength of M-S than Mo-S)~~ and the oxidation might thus be controlled by the reactivity of the underlying Me atom towards oxidation into MOO. Another explanation is that, in the presence of promoter, surface oxidation takes place at low temperatures and the protecting oxysulphide layer thus formed on the edges of MoS, prevents the bulk oxidation of MoS, (which gives rise to the exothermic effect detected by DTA) up to temperatures >600 K. Peuski et ~ 1 . ~ ~ observed the same phenomena for the oxidation of Cogs,. The TPSE experiments show that the amount of H,S released up to 773 K increases in the presence of highly dispersed Co and Ni species on the edges of MoS, crystallites.This suggests that the strength of binding of S on the edges of MoS, slabs is weaker in the promoted catalysts than in pure MoS,.~~S. Gobolos, Q. Wu, 0. AndrP, F. Delannay and B. Delmon 243 1 t 1 I I 500 550 6 00 TR/S IK Fig. 3. Correlation between HDS activity and the onset temperature of reduction/sulphidation of catalyst precursors: x , Mo-HSP-0.0; 0, 0, FeMo; e, D, CoMo; @, 0 , NiMo catalysts with Y = 0.1 or 0.15 (circles) and 0.3 (squares), respectively. XPS One can distinguish schematically four possible locations of the promoter atoms with respect to MoS,: (a) the homogeneous dispersion of promoter and molybdenum sulphides, (b) the segregation of promoter sulphide in the form of a layer covering the surface of the MoS, crystallites, (c) the presence of separate promoter and molybdenum sulphide particles and ( d ) the segregation of MoS, covering the surface of promoter sulphide particles.The XPS intensity data do not provide much insight in enabling us to differentiate between these models. Indeed, the escape length for the 2p electrons can be estimated to be 0.8-1.1 nm using the equation of Szajman and Le~key.~l As shown in table 1, this value is fairly large compared with the size of the elementary catalyst particles. For catalysts prepared by the HSP, IHSP and CP methods, the M/Mo ratios estimated from XPS peak intensities are close to the bulk composition ratios. The only safe conclusion that can be drawn is that complete segregation of either MoS, or MS in the centres of particles is unlikely.In catalysts CoMo-CM-0.1 and CoMo-CM-0.3 the presence of separate phases is established by the appearance of diffraction lines of Cogs, in the XRD spectra. The S 2p peak in the XP spectra is mostly due to sulphur associated with molybdenum in MoS,. This is confirmed by the constant value of AE,(Mo-S) = 66.9-67.0 eV. The significant shift Of AEb(N1-S) and AEb(Co-S) towards higher values as compared with the binding-energy differences in NiS and Co,S,, respectively, suggests various possible phenomena: (i) some electron transfer from Ni or Co to molybdenum, (ii) more ionic Co-S and Ni-S bonds than in Cogs, and NiS, respectively, or (iii) a larger average charge on Ni and Co than in Cogs, and NiS, respectively. These phenomena are not observed in the FeMo sulphide catalysts. Correlation between Catalytic Activity and Physicochemical Properties The correlation between HDS activity and the threshold temperature of temperature- programmed reduction/sulphidation is shown in fig.3. In general, the HDS activity of sulphided catalysts increases with decreasing TR,s. One curve is obtained for the catalysts2432 Activity of MoS,-based Catalysts - ‘WY 10 E N I I I 0 area (arb. units) Fig. 4. Correlation between HDS activity and TPSE spectral area for the promoted catalysts with r = 0.3: 0, FeMo-CP; ., CoMo-CM and HSP; 0, NiMo-HSP and x , Mo-HSP-0.0. prepared by the HSP, IHSP and CP methods; another curve, of similar shape but at a higher temperature, may be drawn for catalysts CoMo-CM-r.Note also that the same curve fits the results obtained with catalysts with r = 0.1 and 0.3. The existence of different curves for the ‘HSP-type’ (HSP, IHSP and CP) and the essentially biphasic CM catalysts1’ could be due to differences in the location and dispersion of the promoter in the two types of samples. This suggests again the importance of highly dispersed M-S species in decreasing TRIs. The existence of the correlation between TRIs and HDS activity may reflect the ‘lability’ of sulphur in both the precursor and the catalyst. It seems that there is a correspondence between the reducibility of the precursors and that of the catalysts or, more precisely, between the reducibility of the precursor and the ease of formation of anion vacancies, which are believed to be the active sites6.8 f 42 in the HDS of thiophene. Correlation between the reducibility of unsupported CoMo sulphide catalysts and their HDS activity has already been reported by Hoodless et al.5 As shown in fig. 4, the present work confirms this recent finding. A linear relationship is found between the HDS activity and the specific TPSE spectral area (referred to unit surface area) for pure MoS, and promoted catalysts with r = 0.3. The TPO and TPSE results listed in table 2 indicate that the higher the reducibility, the lower the oxidisability of the catalysts. Thus, To values also correlate roughly with HDS activity. Both the HDS activity and the threshold temperature of oxidation increase in the sequence Mo-HSP-0.0 < FeMo-CP-0.3 < CoMo-HSP(1HSP)-0.3 < NiMo-HSP- 0.3 for catalysts prepared by methods leading to the formation M-Mo-S structures.(For HSP and CP = ca. 30% M in M-Mo-S was found by Mossbauer spectroscopy in these 42) Oxidisability, which corresponds to a bulk reaction, does not necessarily correlate with oxygen adsorption. The present results should thus, in principle, be considered independently from literature data concerning oxygen chemisorption, which do not seem always to correlate with activity. The differences between the efficiency of promoters in increasing the reducibility, To and HDS activity of MoS, may be related to the binding energy difference between M 2p and S 2p levels. Indeed, another correlation exists between HDS activity and the changeS.Gobolos, Q. Wu, 0. Andre, F. Delannay and B. Delmon 243 3 20- - I N 1 w E AE/eV Fig. 5. Correlation between HDS activity and AE,, = AEb(M-S)catalyst - AEb(M-S)sulfide for the promoted catalyst with r = 0.3; 0, FeMo-CP; ., CoMo-HSP(1HSP) and n, NiMo-HSP) in binding energy with respect to pure Cogs, or NiS. Fig. 5 shows that both the activity and the differences in AE,(M-S) values between MMo-HSP(CP)-0.3 catalysts and MS, increase in the sequence FeMo < CoMo < NiMo. This correlation between binding- energy change, HDS activity and reducibility may be related to electronic effects (the charge on M, the covalency of the M-S bond and electron transfer between M and Mo). If such electronic transfers occur at the surface of promoted catalysts, the M-S bond strength may change and thus also the reducibility and HDS activity of the catalysts.This transfer could be created not only by the formation of Co-Mo-S or Ni-Mo-S structures, but also by a sufficiently close contact between separate Co or Ni sulphide and MoS, phases.16 In the latter case, the different reducibilty and HDS activity of Fe-, Co- and Ni-promoted MoS,-based catalysts could also be related to different abilities of the pure promoter sulphides for producing hydrogen s p i l l ~ v e r . ~ ~ Conclusions The following conclusions may be drawn from the present work. (1) The increase of the HDS activity of unsupported MoS,-based catalysts in the sequence Mo < FeMo < CoMo < NiMo is correlated with a decrease in the threshold temperature of reduction/ sulphidation of the precursors.(2) Both the onset temperature of oxidation of sulphi- dized catalysts and the amount of H,S released in temperature-programmed sulphur extraction increase in parallel with the HDS activity of the promoted catalysts. (3) A change in the binding energy of the XPS Co 2p3,, and Ni 2p3,2 peaks as compared with the pure sulphides, is also correlated with the HDS activity of unsupported catalysts prepared by the HSP(1HSP) and CP methods. (4) A general conclusion is that the reducibility of the precursors and the catalysts plays a major role in determining HDS activity . We are grateful to Mr M. Genet for help with the XPS experiments. S. Gobolos thanks the Central Research Institute for Chemistry of the Hungarian Academy of Sciences and Q.Wu thanks the Beijing Municipal Chemical Industry Research Institute, Cheng Fu, for leaves of absence during which this work was accomplished. The financial support of the Services de Programmation de la Politique Scientifique for a fellowship (S. G.) and general expenditures, and FNRS (Belgium) is gratefully acknowledged.2434 References Activity of MoS,-based Catalysts 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 H. Topsse, B. S. Clausen, R. Candia, C. Wivel and S. Mssup, Bull. SOC. Chim. Belg., 1981, 90, 1190. S. J. Tauster, T. A. Pecoraro and R. R. Chianelli, J. Catal., 1980, 63, 515. N. Y. Topsse and H. Topsse, J. Catal., 1983, 84, 386.G. Hagenbach, P. Menguy and B. Delmon, Bull. SOC. Chim. Belg., 1974, 83, I . R. C. Hoodless, R. B. Moyes and P. B. Wells, Bull. SOC. Chim. Belg., 1984, 93, 613. P. R. Wentrek and H. Wise, J. Catal., 1978, 51, 80. J. T. Richardson, Ind. Eng. Chem., Fundam., 1964,3, 154. J. Matemova, Appl. Catal., 1983, 6, 61. N. K. Nag, D. Fraenkel, J. A. Moulijn and B. C. Cates, J. Catal., 1980, 66, 162. C. G. Gachet, E. Dhainaut and L. de Mourges, Prepr. Petr. Chem. Diz;. ACS, 1982, 27, 753. C. Gachet, R. Paulus, L. de Mourges, C. Durand and H. Toulhoat, BUN. SOC. Chim. Belg., 1984, 93, 681. R. Candia, B. S. Clausen and H. Topsse, Bull. SOC. Chim. Belg., 1981, 90, 1225. M. Breysse, R. Frety and M. Vrinat, Prepr. Petr. Chem. Dic. ACS, 1982, 27, 772. G. Hagenbach, Ph.Courty and B. Delmon, J. Catal., 1971, 23, 295; 1973, 31, 264. S. Gobolos, Q. Wu and B. Delmon, Appl. Catal., 1984, 13, 89. S. Gobolos, Q. Wu, J. Ladriere, F. Delannay and B. Delmon, Bull. SOC. Chim. Belg., 1984, 93, 687. Q. Wu, S. Gobolos, P. Grange and F. Delannay, Thermochim. Acta, 1984, 81, 281. C. D. Wagner, L. E. Davis, M. V. Zeller, J. A. Taylor, R. H. Raymond and L. H. Gale, Surf. Interface Anal., 1981, 3, 21 1. Q. Wu, S. Gobolos, F. Delannay and B. Delmon, in 10th Symp. Reactivity of Solids, ed. P. Barret and L-C. Dufour (Elsevier, Amsterdam, 1985), p. 1067. V. I. Zaikovskii, A. P. Shepelin, V. A. Burmistrov, A. N. Startsev and Yu. I. Yemakov, React. Kinet. Catal. Lett., 1984, 25, 17. C. K. Groot, A. M. van der Kraan, V. H. J. de Beer and R. Prins, Bull. SOC. Chim. Belg., 1984,93,707. B. S. Clausen, H. Topsse, R. Candia and B. Lengeler, ‘Catalytic Materials: Relationship between Structure and Reactivity’, ACS Symp. Ser., 1983. F. Delannay, D. S. Thakur and B. Delmon, J. Less-Common Met., 1979, 63, 265. D. S. Thakur, P. Grange and B. Delmon, J. Less-Common Met., 1979, 64, 201. E. Furimsky and C. H. Amberg, Can. J. Chem., 1975, 53, 3567. N. M. Zaidman, D. B. Osechkin, M. F. Gladovskaya and E. N. Martynova, Khim. Tekhnol. Topl. Masel, 1961, 6, 25. Ph. C. H. Mitchell and C. E. Scott, Bull. SOC. Chim. Belg., 1984, 93, 619. M. Vrinat, M. Breysse and R. Frety, Appl. Catal., 1984, 12, 151. M. Breysse, R. Frety and M. Vrinat, Appl. Catal., 1984, 12, 165. H. Topsse, B. S. Clausen, R. Candia, C. Wise1 and S. Msrup, J . Catal., 1981, 68, 433. N.-Y. Topsse, H. Topsse, 0. Ssrensen, B. S. Clausen and R. Candia, Bull. SOC. Chim. Belg., 1984,93, 727. 0. Ssrensen, B. S. Clausen, R. Candia and H. Topsse, Appl. Catal., 1985, 13, 363. C. Wivel, R. Candia, B. S. Clausen and H. Topsse, J. Catal., 1981, 68, 453. H. Topsse, R. Candia, H-Y. Topsse and B. S. Clausen, Bull. SOC. Chim. Belg., 1984, 93, 783. R. R. Chianelli, T. A. Pecoraro, T. R. Halbert, W-H. Pau and E. I. Stiefel, J. Catal., 1984, 86, 226. S. Msrup, B. S. Clausen and H. Topsse, J. Phys. Colloy., 1979, 40, C2-88. S. Gobolos, Q. Wu, J. Ladrikre, F. Delannay, P. Grange and B. Delmon, to be published. E. Ya. Rode and B. A. Lebedev, Russ. J. Inorg. Chem., 1961, 6, 608. 0. P. Bahl, E. L. Evans and J. M. Thomas, Proc. R. SOC. London, Ser. A, 1968, 306, 53. A. V. Peuski, A. R. Bobenko and R. G. Kefer, Fro. Vyssh. Uchebn. Zaved. Tswtn. Metall., 1973, I, 37. J. Szajman and R. C. G. Leckey, J. Electron Spectros., 1981, 23, 83. G. C. A. Schuit, Int. J. Quantum Chem., Suppl., 1977, 12(2), 43. Paper 5 / 1624; Received 19th September, 1985