J. CHEM. SOC. FARADAY TRANS., 1994, 90(14), 2125-2131 Surface Distribution and Heteroatom Removal Activity of Equilibrium Adsorption Prepared, Doubly Promoted (Zn,Co)Mo/Al,O, Catalysts H. Thomas, C. Caceres and M. Blanco Centro de lnvestigacion y Desarrollo en Procesos Cataliticos (CINDECA), CONICET-UNLP,Calle 47 No. 257, 1900-La Plata ,Argentina J. L. G. Fierro and A. Lopez Agudo* lnstituto de Catalisis y Petroleoquimica , CSIC, Campus de la UAM, Cantoblanco, 28049-Madrid, Spain Adsorption isotherms of Zn2+ and/or Co2+ ions on an 8 wt.% MoO,/alumina catalyst have been obtained. The Langmuir parameters derived from regression of the results indicate that preimpregnation of alumina with molybdate solutions notably decreased the adsorption of Zn2+ and Coz+ ions.In solutions containing both Znz+ and Coz+ ions, the saturation capacity of Zn2+ was practically unaffected, while that of Co2+ was significantly reduced owing to the higher intrinsic affinity of Zn2+ ions for alumina sites. Adsorption of Co2+ ions occurs on two type of sites, i.e. those on which Co2+ ions are loosely and strongly bound; the Co2+ on the latter type is not affected by the presence of Zn2+ ions. After calcination, the non-adsorbed Co, retained in the pores, was depos- ited heterogeneously as a surface cobalt oxide bound loosely to the A120,. However, the homologous Zn was more homogeneously deposited, owing to its higher reactivity with A120,. These differences result in an addi-tional promotional effect in the gas-oil hydrodesulfurization (HDS) reaction, which has been observed on the catalysts prepared by simultaneous incorporation of Zn and Co.This effect is explained in terms of changes in the distribution and dispersion of promoters. These changes appear to be detrimental in hydrodenitrogenation (HDN). The more commonly used catalysts in hydrotreating pro- cesses consist of molybdenum sulfide as the active phase, sup- ported on a y-alumina carrier. The incorporation of a second metal promoter, such as Ni, Fe, Zn or Mg, has often been practised to improve the HDS activity.'-4 Several studies of molybdenum catalysts have been reported in which the cata- lytic effect of Zn addition was found to be owing mainly to differences in preparation procedures involving catalyst composition and Zn loading.In a systematic study of a series of CoZn-Mo/A1203 catalysts in which both Co and Zn were simultaneously impregnated on an Mo/A1203 sample. Fierro et a1.' found a secondary promoting effect for gas-oil HDS when Co was partially replaced by Zn. This effect of Zn was attributed to a decrease in the formation of the catalytically inactive CoA1204 phase at the alumina inter- phase and, consequently, to an improvement in the amount of octahedrally coordinated surface Co, which was induced by the preferential coordination of Zn2 + to tetrahedral sites, rather than Co2+. This is consistent with the higher propen- sity of Zn than Co to form a surface spinel with alumina, as reported by Lo Jacono and Schiavello" and by Strohmeir and Hercules." On the other hand, Maezawa et a1." have shown that Zn did not significantly change the Mo dispersion on Zn-Mo/A1203 ,and that Mo could affect the Zn distribu- tion, depending on the order of impregnation of the Mo and Zn components.In order to clarify the origin of the observed secondary promotional effect of Zn on Co-Mo/Al,O, catalysts, we decided to examine in more detail the first step of the catalyst preparation. By studying the separate and simultaneous adsorptions of the different metal components on the alumina surface, it should be possible to identify the ion that is preferentially adsorbed and its maximum coverage. Washing of the samples, after ion adsorption, can give infor- mation on the strength of the adsorbate-support interaction and the possible presence of different types of adsorption sites.Moreover, characterization of the calcined forms of the precursors should reveal the chemical interactions taking place between the components in the calcination step and changes in the metal dispersion. In previous work,13 the adsorption isotherms of the simpler systems, Co/A1203, Zn/A120, and ZnCo/A1203 , were studied, and the resulting calcined samples were charac- terized. This study revealed that Zn2+ species are prefer- entially and more strongly adsorbed on an alumina surface than Co2 species, which leads to important differences in + the distributions of Zn and Co on the alumina surface upon calcination. Information on the adsorption and interaction of molybdates on different supports, but particularly on alumina, has been reported in many st~dies.'~-'~ Recently, it has been shown that coadsorption of Mo6+ species and Co2+ ions on the surface of y-alumina enhances the extent of adsorption of both Co2+ ions and Mo6+ species.23 Since the promoters are often impregnated in a second step, after molybdenum, a study of the adsorption of cobalt and zinc on a calcined Mo/A1203 sample is of great interest.Preliminary results of the equilibrium adsorption isotherms of Zn2+ and/or Co2 + ions on a molybdate-modified alumina support were recently presented.24 Here, a complete data analysis of the above-monitored adsorption isotherms and the results of an extensive characterization of a selection of representative calcined samples will be presented.Also, to corroborate the promoting effect of Zn on HDS activity' and to examine if such effects also occur in other hydrotreating reactions, addi- tional activity results for the simultaneous HDS and HDN of gas oil + pyridine on samples of Zn-, Co-and ZnCo-Mo/ A120, catalysts prepared by equilibrium adsorption (similar to the samples used for the adsorption isotherms) will be pre- sented and discussed. Experimental Sample Preparation The alumina support used was a commercial Girdler T-126 y-A1203(S,,, , 188 m2 g- ;pore volume, 0.4cm3 g- ') which was crushed and sieved to a particle size of 0.84-1.19 mm. An Mo/Al,03 sample was prepared by equilibrium adsorption of an ammonium heptamolybdate solution (20 g Mo 1-') on alumina.After impregnation, the solid was fil-tered, dried at room temperature for 24 h and then calcined in air at 823 K for 7 h. The Mo loading in the solid (8 wt.% MOO,) was determined by treating the sample with a strong acid solution and analysing the Mo in the solution by atomic absorption spectroscopy (AAS). The equilibrium adsorption experiments of Zn2+ and/or Co2+ ions were performed at 293 K as follows: 1 g of the above-prepared Mo/Al,O, sample was suspended in 4 ml of aqueous Zn solution and/or Co nitrate and stirred contin-uously for 24 h. The concentration of the solutions was varied from 1 to 50 g of active metal (Zn or Co) per 1. The Co :Zn molar ratio for all mixed solutions was 1:1.After equilibration, the solids were separated from the solution by centrifugation, and dried and calcined under the same condi-tions as the precursor Mo/Al,O, sample. The concentration of Zn and/or Co in the initial (Ci) and the equilibrium or final (C,) solutions was determined by measuring the metal content in such solutions by AAS. From Ci and C, the concentration of metal adsorbed (C,) on the Mo/Al,O, sample was calculated by mass balance and by assuming that the pore volume of Al,O, did not vary during adsorption. Details of the general procedure of adsorption and calculations of C, have been given else~here.'~*~~ To study the strength of adsorption of the ions on the support, wet samples (1 g) of the equilibrium adsorption experiments were washed repeatedly with 4 ml of deionized water for several days.The amounts of Zn and Co removed in each washing were determined by AAS. Three additional samples, Zn-, Co-and ZnCo-Mo/Al,O, , used for the catalytic activity measurements, were also pre-pared by equilibrium adsorption following the general pro-cedure described above. Specifically, the single promoted Zn-and Cc~Mo/A1,0, catalysts were prepared by suspending 7 g of MOO,@ wt.%)/Al,O, in 28 ml of an aqueous solution of zinc nitrate (22.1 g Zn 1-') or cobalt nitrate (20.2 g Co 1-'), respectively. The doubly promoted ZnCo-Mo/Al,O, catalyst was prepared in a similar manner using a solution containing 10.1 g Co 1-' and 11.1 g Zn 1-l.The separation of the solids and the calcination conditions were as mentioned above for the precursor Mo/Al,O, sample. The concentrations of adsorbed metal, both occluded in the pores and the total amount in the calcined catalysts, are summarized in Table 1. CharacterizationTechniques Diffuse reflectance spectra (DRS) were recorded on a Varian Super Scan 3 spectrophotometer using BaSO, as reference. X-Ray photoelectron spectra (XPS)were recorded on a Leybold Heraeus LHS 10 spectrometer equipped with an A1 anode. Each spectral region was signal-averaged for a number of scans to obtain good signal-to-noise ratios. The Table 1 Concentration of metal (wt.%) in the catalysts prepared by equilibrium adsorption catalyst C-Mo/AI 20, Zn-Mo/Al,O, ZnCc~Mo/Al,0, co Mo Zn Mo Zn Co Mo C, 0.45 -0.23 -0.25 0.13 -CP" 1.64 -2.35 -1.09 0.93 -C, 2.09 5.36 2.58 5.36 1.34 1.06 5.36 "Occluded in pores of the support.J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 C ls,O ls,Al2p,Mo3d,Co2pand/orZn2pwererecordedforeach sample.Allbindingenergies(E,,)werereferencedtotheA12ppeakat 74.7. XPS intensity ratios were determined using the integrated areas of the Co 2p(and satellite),Zn 2p and A12p lines afterlinear background subtraction. Additional experiments involving chemical extraction af Co and Zn species were conducted by contacting lo0 mg of sample with 25 ml of a 3% (v/v) ammonia solution for 24 h. The metal content in the solution was determined by AAS. Activity Measurements Catalyst activity testing for the simultaneous HDS and HDN of gas oil and pyridine was carried out in a trickle-bed flow reactor at 30 kg ern-,, with a liquid hourly space velocity (uLHS)of 8.8 h-',and an H,(gas) :feed(1iquid) ratio of 408 :1, at 598, 623 and 648 K.A catalyst charge of 6 ml was diluted 1:1 with Sic of the same particle size. The catalysts were presulfided in situ at 623 K for 25 h with a 7% (v/v) CS,-gas oil mixture at 20 bar and a VLHs of 2. Sulfur and nitrogen contents were simultaneously determined with an Antek analyser system. From the percentages of sulfur and nitrogen removed, the activities of the catalysts for HDS and HDN, expressed as pseudo-second-order (kHDs) and pseudo-first-order (kHDN) rate constants, were calculated using the equa-tions : where xs and xN are the sulfur and nitrogen conversions, respectively.Results Adsorption Isotherms Fig. 1 shows the adsorption isotherms of Zn2+, Coz+ and (Zn2++Co2+) on MoO,/Al,O, at 293 K. They roughly obey the Langmuir isotherm. The solid lines of Fig. 1 are best fits of the experimental data to the Langmuir model. As in the adsorption on al~mina,'~the shapes of the adsorption Oa50 (c)t U n-0 c 0.25k+cb+ J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 isotherms for Zn2+ are slightly different from those for Co2+. The amount of metal adsorbed, Ca, increased and rapidly levelled off for Zn2+ adsorption, while C02 adsorption was + more gradual. The experimental results were fitted to the Langmuir iso- therm: (3) where K, is the adsorption equilibrium constant and C, the number of adsorption sites expressed in g metal per 100 g of support. From eqn.(3), the values of K, and C, for each isotherm were calculated; they are shown in Table 2. The K, values for Zn2+ are significantly higher than those for Co2+, indicating that Zn2+ ions have a higher affinity for alumina than Co2+ ions. The higher affinity of Zn2+ is related to the higher stability and preferential adsorption of the hydrolysed Zn2+ complex over the corresponding Co2+ 0ne.13 Also, K, for Zn2+ (for the coadsorption of Zn2+ and Co2+) is lower than that for the adsorption of Zn2+ alone, indicating that the affinity of Zn2+ for alumina is significantly decreased by the presence of Co2+ ions because both ions are competing for the same type of adsorption site.However, K, for Co2+ (during coadsorption) is higher than that for the adsorption of Co2+ alone, as was also found for adsorption on Al,O, without molybdena.13 In this paper we suggested that the majority of Co2+ ions may be adsorbed on the same sites as Zn", and that the remaining ions can be found on sites with stronger affinity, on which Zn2+ ions are not adsorbed.', Thus, the K, value of Co2+ for the coadsorption of Zn2+ and Co2+ may refer to the irreversible adsorption of Co2+ on the stronger sites. In line with this, Table 2 shows that the C, values of Zn2+ adsorption are practically equal for simulta- neous and monocationic adsorption, while the Cm for Co2+ is clearly smaller for the simultaneous adsorption case, since Zn2+ ions have a higher affinity for alumina than Co2+ ions.Moreover, the C, for Co2+ alone was clearly higher than that for Zn2+ alone, which is consistent with the presence of two distinct types of Co2 + adsorption site. Metal Leaching from Wet Samples The variations in Co and Zn content as a function of washing time for some representative samples of the adsorption iso- therms are shown in Fig. 2. The number in parentheses in the catalyst notation refers to the sequence order of the corre- sponding isotherm in Fig. 1. For both low- and high-Co- content Co-Mo/Al,O, samples the initial amounts of adsorbed Co, i.e. C, , decreased markedly with increasing washing time, and after about 75 h of washing the Co content remained practically constant. This means that most of the adsorbed Co2+ ions on Co-Mo/Al,O, samples are loosely bound species, which can be easily removed by repeated washing with water, and that only a small part of the Co species are strongly bound to the support, remaining adsorbed.However, Fig. 2A shows that for Zn,Co-Mo/Al,O, samples practically no extraction of Co Table 2 Equilibrium adsorption parameters system KJ(g Me)-' * C,/g(lOO g Mo/Al,O,)-' Co-Mo/Al 203 Zn-Mo/Al,O, 0.10 4.58 0.64 0.24 ZnCo-Mo/Al,O, 0.22 1.30 0.19 0.27 Me: metal. 2127 4 t - I I I 25 50 75 335 time of washing/h Fig. 2 Variation in A, Co and B, Zn content as a function of washing time for different wet samples of the catalysts in Fig.1. A, (a) ZnCo-Mo (10);(b) Co-Mo (10);(c) ZnCo-Mo (2), (d)Co-Mo (5). B, (a)Zn-Mo (9),(b) ZnCo-Mo (lo),(c)Zn-Mo (2), (d) Zn-O-Mo) (2). was observed. Therefore, in these samples only Co species strongly bound to the support were present. Notice also from Fig. 2A that the Co contents of the Co-Mo/Al,O, samples after about 14 days of repeated washing were approximately equal to the amounts of Co initially adsorbed in the counter- part ZnCo-Mo/Al,O, isotherm sample. A similar behaviour of the adsorbed Co species upon washing was also found for Co/Al,O, and ZnCo/Al,O, samples in the previous study.' The effect of washing on adsorbed Zn, if any, was very small for both Zn- and ZnCo-Mo/Al,O, samples, as Fig. 2B shows.The Zn content after repeated washings was equal to or slightly lower than its initial value. Hence, this behaviour indicates that all adsorbed Zn species are irreversibly bound, involving only one type of site. Diffuse Reflectance Spectra Diffuse reflectance spectra of the calcined Co-Mo/Al,O, and ZnCo-Mo/Al,O, samples show triplet bands at ca. 545, 580 and 630 nm, characteristic of tetrahedrally coordinated Co2 + (CoCT]) in the surface spinel CO/A~,O,.~~ The band at ca. 750 nm, due to octahedrally coordinated Co2+ (Co[O]) in CO~O,,~~was detected in samples with Co loadings above 0.15 wt.% for the ZnCo-Mo/Al,O, series, and at much higher contents, 3.5 wt.% Co, for the Co-Mo/Al,O, series.Fig. 3 shows that the ratio of Co[T] to Co[O], as determined by the Schuster-Kubelka-Munk function F(R,) ratio mea- sured at 580 and 750 nm, remained practically constant up to C, level of about 2 wt.% Co in the Co-Mo/Al,O, samples and then abruptly decreased to a very low value for samples with loadings >3.5 wt.% Co.This variation was reversed for the ZnCo-Mo/Al,O, samples, in which the ratio of Co[T] to CoCO] initially decreased very rapidly for contents up to about 0.3-0.5 wt.% Co and then became nearly constant or gradually decreased. Note that the ZnCo-Mo/Al,O, samples had a much lower Co[T] : Co[O] ratio, reflecting the lower C,/g Co (1 00 g AI,O,) -' Fig. 3 Variation of the Kubelka-Munk function ratio at 580 and 750 nm with Co content for calcined C+Mo/Al,O, (0)and CoZn-Mo/Al,O, (0)samples extent of adsorption of Co in the presence of Zn and, there- fore, the small amount of Co available for strong interaction with the alumina.Zinc Extraction Table 3 shows the effect of calcination temperature on Zn extraction by aqueous ammonia solution from four Zn-Mo/Al,O, and ZnCo-Mo/Al,O, representative samples, selected in the low and high regions of the corresponding adsorption isotherms. It is observed that for low-Zn-content samples the amounts of Zn removed were very low and did not vary significantly when the calcination temperature increased from 623 to 823 K, indicating that the fraction of Zn as ZnO (very soluble in ammonia) was very small.In such samples the percentage of Zn extracted was around 1%, which corresponds to the occluded Zn in pores being trans- formed upon calcination to extractable ZnO. However, for the high-Zn-content samples calcined at 623 K the percent- ages of Zn extracted were relatively high (15.7 and 5.5%), decreasing notably to ca. 1 and 0.3% upon calcination at 823 K. This decrease in Zn extraction with increasing calcination temperature means that a significant proportion of the Zn, present as ZnO in the dried samples, was transformed to a ZnAl,O, surface spinel which is only slightly solubilized by ammonia. X-Ray Photoelectron Spectroscopy The XPS data of the most representative calcined samples (used for the isotherms) are presented in Table 4.It is observed that the binding energies (Eb) of the A1 2p, Co 2p3,,, Zn 2p,,? and Mo 3d,,, lines did not vary to any significant extent, irrespective of the metal content and extent Table 3 Influence of calcination temperature on the extent of Zn extraction sample calcination temperature/K C," Ce" Zn-Mo (3) 623 2.7 0.004 823 2.7 0.03 Zn-Mo (12) 623 823 17.2 17.2 2.7 0.2 ZnCo-Mo (2) 623 1.4 0.012 823 1.4 0.016 ZnCo-Mo (12) 623 823 23.5 23.5 1.3 0.08 Zn concentration is given in mg per g support; t, total; e, extracted. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 4 Binding energies (eV) of core electrons in Co-, Zn-and CoZn-Mo/Al,O, catalysts catalyst 0 1s Mo 3d,,, Co 2p3,, Zn 2P3,~ 8 Mo/Al 532.1 233.3 - - CoZn-Mo (3) CoZn-Mo (5) CoZn-Mo (6) CoZn-Mo (8) CoZn-Mo (9) CoZn-Mo (1 1) Co-Mo (4) Co Mo (5) Co-Mo (6) Co-Mo (8) Co-Mo (10) Zn--Mo (1) 531.9 531.7 531.9 531.8 532.0 532.0 531.7 532.0 531.9 531.9 532.8 53 1.9 233.0 233.1 233.3 233.2 233.0 233.0 233.0 232.8 232.7 232.8 232.9 232.9 780.4 780.8 780.8 780.6 780.3 780.4 780.9 780.8 780.9 780.9 780.6 1022.7 1022.7 1023.2 1022.8 1023.4 1022.7 1022.7 Zn-Mo (4) Zn-Mo (6) Zn-Mo (10) Zn-Mo (1 1) Zn-Mo (12) 531.9 532.0 531.8 531.9 53 1.9 232.9 233.0 232.9 233.0 233.0 1022.8 1022.7 1022.6 1022.7 1022.8 of adsorption, indicating, in principle, the presence of similar metal species in all calcined samples.The Ebs of co 2p3,, and Zn 2p,,, at ca. 780.6 and ca. 1022.6 eV, respectively, are very close to those of the CoAl,O, and ZnAl,O, surface spinels.' Quantitative XPS data of the Zn and Co dispersions are presented in Fig.4 as a function of the bulk metal content. The experimental intensity ratio, Izn/IA,, of the Zn-Mo/Al,O, samples increased linearly up to a Zn content of ca. 1.3 wt.%, where an inflection was observed, indicating a homogeneous deposition of Zn on the support up to the mentioned Zn content, and, beyond such a level, a change and decrease in the Zn dispersion. For the ZnCo-Mo/Al,O, samples, the change in the linearity of Iz,/IA, and, therefore, in Zn dispersion appeared at about 0.6 wt.% Zn. Another difference is that the ZnCeMo/Al,O, samples exhibited higher Izn/IA,ratios than the Zn-Mo/Al,O, counterparts, which were also greater than the theoretical intensity ratios for a monolayer distribution.This suggests that the oxidic Zn species of the ZnCO-Mo/Al,O, samples are more inhomoge- neously distributed, yielding some enrichment of Zn in the 4 6 8 10 12 10, (Me/Aub",k Fig. 4 Experimental XPS intensity ratios of (a) Z,,/ZA, (0,m)and (b) I,-,/IAl (0,0)as a function of the bulk metal ratio (chemical). Open symbols for Co- and Zn-Mo/Al,O, samples and full symbols for ZnCo-Mo/Al,O, samples. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 I O' 014 OI8 112 1f6 210 C,Jg Me (1 00 g AI,O,) -' Fig. 5 Experimental XPS intensity ratio (ZMJZA,) as a function of the total metal (Co, Zn or Co + Zn) content in the calcined samples: (A) Zn-Mo/Al,O,, (0)CCFMO/AI,O, and (0)ZnCc+Mo/Al,O, outer part of the support grains.The variation of Ico/I,, with bulk Co content are shown in Fig. 4. The trends of these curves are, in general, similar to but the same as those for Zn, except that the deviation from linearity occurred at a much lower metal content, i.e. at about 0.7 wt.% Co for the Co-Mo/Al,O, samples and at about 0.3 wt.% Co for the ZnCo-Mo/Al,O, samples, indicating that Co became distrib- uted heterogeneously at much lower metal contents than Zn, as was also found for the samples without Mo.', Fig. 5 shows that the IM,,/IAlratios are practically the same for all samples, indicating that the dispersion of Mo was not significantly altered by the later adsorption of increasing amounts of Zn and/or Co.A similar finding was reported for catalysts prepared by using conventional impregnation methods. l2 This figure also shows that the parent Mo/A1,0, sample had an IMo/I,,ratio slightly higher than those of the Co-and Zn-containing samples. This is probably due to a dilution effect and/or a somewhat partial covering of the Mo with metal promoters. Catalytic Activity The activities of the three catalysts prepared by equilibrium adsorption for simultaneous HDS and HDN of gas oil and I f ;3 a 2 1 0 598 623 648 N' nt22 I -L 1 0 598 623 648 Fig. 6 Catalytic activity of Zn-, Co-and ZnCo-Mo/Al,O, cata-lysts for HDS and HDN of gas oil + pyridine at 598,623 and 648 K pyridine are shown in Fig.6. It is evident that the ZnCo doubly promoted catalyst exhibited higher HDS activity than the Zn-promoted catalyst, and the latter was more active than the Co-promoted catalyst; whereas for HDN, the Zn- promoted catalyst proved to be notably more active than the ZnCo doubly promoted catalyst and also much more active than the Co-promoted one. Discussion Influence of Preimpregnated Molybdena on Zn and/or Co Adsorption on the Surface of y-Alumina Comparison of the results in Table 2 with those previously obtained for adsorption on alumina', reveals that the afin- ities of both Zn2+ and Co2+ions for alumina sites decreased significantly (by at least a factor of two) after molybdena incorporation. Furthermore, all of the C, values of Table 2 (except for Co2+ in the coadsorption case, see below) for Mo/Al,O, are lower than those obtained on alumina alone.', In the present adsorption systems the Zn2+ and/or Co2+ ions are not coadsorbed with another negatively charged species, at variance with ref.23, because the molyb- date species were previously impregnated on the alumina and then calcined. This decrease in Co and Zn adsorption due to Mo incorporation cannot be explained if an electrostatic mechanism is assumed to describe the adsorption of M ions onto the alumina surface. As pointed out by Br~nelle,~~ the extent of cation adsorption is favoured when the pH of the solutions is higher than the isoelectric point (IEP) of the carrier, and this should be expected since the modification of alumina with 8% MOO, caused a decrease in the number of its basic hydroxy group^,'^-^' reflected by a decrease of cu.2 pH units in the IEP of alumina.28 This effect would cause the IEP of alumina to approach the pH of Zn2+ and Co2+ solu- tions, increasing the probability of having a negatively charged surface, and consequently it should favour the adsorption of cations, as also occurred upon doping the y-alumina with F-ions2, However, there is another phenome- non with the reverse effect. From electrophoretic migration rates, ca. 80% of the alumina surface was covered by molyb- date when the adsorption was allowed to take place,28 thus reducing the concentration of acidic hydroxy groups avail- able (for adsorption of metal ions) during calcination.The lower C, values in Table 2 relative to that for alumina alone', suggest that a fraction of molybdate is adsorbed on the same types of sites that are involved in adsorption of Zn2+ and most Co2+ ions. An alternative explanation is that molybdate prevents, probably by steric effects, the interaction between surface sites and the large aqueous metal ions, as has already been suggested for molybdate adsorption on phos- phated alumina.,' The exception in the C, values for Co2+ in the coadsorption of Co2+ and Zn2+, mentioned above, (these C, values were very similar for both ZnCo/Al,O, C0.16 g (100 g support)-'] and ZnCo-Mo/Al,O, C0.19 g (100 g support)-'] systems], is consistent with the previous assumption that a small fraction of Co2+ ions are probably adsorbed on differ- ent sites to those involved in Zn2+ adsorption. This conclu- sion is corroborated by the above results (Fig.2) of Co removed, from wet Co-Mo/A1203 samples, by washing with water. Such results confirm the presence of two types of alumina site (whose nature cannot be clarified from the present results) for the adsorption of Co2+ ions. In the case of coadsorption of Zn2+ and Co2+, the latter ions occupy only the sites on which they are strongly adsorbed. By con- trast, the Zn2+ ions are adsorbed very strongly on only one type of site. Dispersion and Distribution of Zn and/or Co in the Calcined Samples Upon calcination at 823 K of the Zn-, C* and ZnCo-Mo/Al,O, samples, as also occurred for samples that had been employed for adsorption on non-modified Al,O, ,13 the adsorbed and occluded Zn and/or Co were mostly incorporated into the defective lattice of Al,O, ,occupying, in general, tetrahedral and octahedral sites at low and high metal contents, respectively.Although the previous incorpor- ation of Mo into Al,O, considerably decreased the adsorp- tion of Zn and/or Co, compared with that on Al,O,, the segregation of bulk crystalline ZnO and Co,O, was not detected by X-ray diffraction in the present calcined samples. In agreement with this, the XPS E,s of the Zn 2p3,, and Co 2p,,, levels, which remained practically invariable for all samples, indicated that Zn and Co were present as ZnAl,O, and CoAl,O, surface spinels, respectively, rather than as ZnO and c030, structures.However, the observed changes in the linearity of the intensity ratios of Zn 2p3,, and Co 2p3,, to the A1 2p levels us. metal content (Fig. 4) clearly indicated that the surface distribution and dispersion of Zn and Co changed at certain metal loadings depending on the extent of adsorption. For the Zn-Mo/Al,O, samples, the change in Zn distribution occurred at about 1.3 wt.% Zn, which is slightly over the maximum Zn loading value of 0.24 wt.% obtained from the adsorption isotherm. This difference indicates that below 1.3 wt.% Zn most of the occluded Zn in the pores interacted with the alumina during the calcination step and formed similar ZnAl,04-like species to those resulting from adsorbed Zn, and that at higher Zn loadings not all of the occluded Zn can interact strongly with the alumina because the latter is partially covered by Mo and, therefore, some Zn remains as highly dispersed ZnO.In the ZnCo-Mo/Al,O, samples, this change in the homogeneous distribution of Zn appeared at lower Zn loadings (ca. 0.6 wt.%) because the percentage of occluded Zn was slightly higher than that for the Zn-Mo/Al,O, samples and also because of the presence of adsorbed and occluded Co, which can interact with alumina, although with lower reactivity than Zn. This phenomenon of the spreading of ZnO species, arising from occluded Zn, is supported by the Zn extraction results at different calcination temperatures for the different Zn-containing samples shown in Table 3.The change in the homogeneous distribution of Co, according to the XPS data (Fig. 4), occurred at Co contents of about 0.7 and 0.3 wt.% for the Co-Mo/Al,O, and ZnC*Mo/Al,O, samples, respectively. Such values were very close to the saturation coverages calculated from the adsorption isotherms, C, values are given in Table 2. This finding indicated that there was no significant spreading of the Co oxide species derived from occluded Co, owing to the relatively minor reactivity of the Co oxide species with alumina which may remain highly dispersed on the support, particularly for the Co-Mo/Al,O, samples with up to ca. 2 wt.% Co, on the basis of the DRS results.By comparing these results with the corresponding XPS and DRS results on Co/Al,O, samples,13 it is observed that the change in the slope of the XPS results and the Co[T] : Co[O] ratio appeared at 'lightly higher c0 loadings for the Co-Mo/A1203 than for the Co/A1203 Thisis consistent with the known fact that Mo improves the dis- persion of Co over A1,0, .12*30,31 Promotional Effect of Zn The activity results of Fig. 6(a) confirm the earlier findingg of an additional Promotion in HDS activity when the Zn and Co promoters were simultaneously incorporated into an J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Mo/Al,O, sample. The levels of activity of the present cata- lysts were, however, lower than those of the previous studyg in which the catalysts were prepared by conventional impreg- nation methods, but in both cases the relative promoting effect was similar.The increased HDS activity of the ZnCo-Mo/Al,O, cata-lyst, in comparison to the Zn- and Co-Mo/Al,O, catalysts, cannot be attributed to changes in the Mo dispersion since this remained essentially unchanged for all Co- and Zn- containing samples. The increase in HDS activity is, there- fore, associated with the observed changes in the distribution and dispersion of the promoters. According to Table 1, the doubly promoted ZnCo-Mo/Al,O, catalyst showed a lower proportion of adsorbed Co and a higher proportion of occluded Co than the singly promoted Co-Mo/Al,O, cata-lyst. This may lead, after calcination, to a higher proportion of Co[O] weakly bound to the alumina in the former cata- lyst, as the above characterization results indicate.Therefore, is seems that such Co[O] species, presumably in the form of Co oxide highly dispersed on the alumina surface, can form more easily a more abundant Co-promoted molybdenum active phase. The fact that the Zn-Mo/Al,O, catalyst was somewhat more active for gas-oil HDS than Co-Mo/Al,O, can, in prin- ciple, also be attributed to the lower C& ratio of the pro- moter in the former catalyst (Table l),which may lead to a higher proportion of promoter that is not strongly bound. In addition, note that the two catalysts differ in the total content of promoter, the wt.% of Zn was slightly higher than that of Co.The data of de Beer et aL3, also showed that at certain promoter : Mo ratios, the promoter action of Zn for thio- phene HDS was significantly higher than that of Co, Ni and Mn. As shown in Fig. qb), for HDN of pyridine, Zn-Mo/Al,O, was considerably more active than Co-Mo/Al,O, , which can be attributed to the comparatively higher hydrogenation activity of Zn respect to Co3, and also to the above- mentioned differences in promoter distribution. A similar large difference in HDN activity between Zn and Co was also found for molybdenum sepiolite-based catalyst^.,^ In this case, the higher surface acidity exhibited by the Zn-promoted catalyst relative to the Co-promoted one was suggested as another factor enhancing the HDN of the Zn-promoter cata- lyst.This could also occur with the present catalysts. From Fig. 6(b) it is also evident that in the ZnC*Mo/Al,O, catalyst no additional promotional effect for HDN activity appeared, in contrast to what is observed for HDS, since its HDN activity was below the sum of the corresponding activities of the Zn- and Co-promoted cata- lysts. This behaviour for HDN is consistent with the intrinsic activity of the singly promoted catalysts and the surface dis- tributions of both Zn and co. In the ZnCo-Mo/Al,O, cata-lyst most of the Zn is strongly bound to the support and, therefore, cannot contribute very much to the HDN reaction, whereas most of the Co is weakly bound to the support and is thus able to form active sites with Mo, but the latter are less active for HDN.Consequently, the HDN activity of this catalyst is relatively low. Financial support from the DGICYT (Spain), Project PB 87-0261, is gratefully acknowledged. The technical assistance of Mr. osiglio is also gratefully acknowledged. References 1 0.Weisser and S. Landa, Sulphide Catalysts, Their Properties and Applications, Pergamon, New York, 1973. 2 P. C. H. Mitchell, in Catalysis (ed. C. Kemball), The Chemical Society,London, 1977, vol. 1, p. 228. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 2131 3 4 5 A. L. Hensley, US Pat., 3 849 296, 1974. H. Beuther, R. A. Flinn and J. B. McKinley, Ind. Eng. Chem., 1959,51, 1349. F. J. Lopez, H. Medina, N. P. Martinez and P. C. Mitchell, 21 22 N. Spanos, L.Vordonis, Ch. Kordulis and A. Lycourchiotis, J. Catal. 1990, 124, 301. P. G. Koutsoukos and A. S. Lycourghiotis, Colloids Su$, 1991, 55, 297. 6 React. Kinet. Catal. Lett., 1983, 1, 2, 153. F. Trifiro, P. L. Villa, P. Pasquon, A. Iannibello and V. Verti, 23 24 N. Spanos and A. Lycourghiotis, Langmuir, 1993,9,2250. H. Thomas, C. Caceres, M. Blanco, J. L. G. Fierro y A. Lopez 7 Chim. Ind. (Milan),1975,57, 173. G. Muralidhar, F. E. Massoth and J. Shabtai, J. Catal., 1984,85, 44. 25 Agudo, in Actas del XIZI Simposio lberoamericano de Catalisis, Segouia, Spain, 1992, vol. 1, p. 499. M. A. Aulman, G. J. Siri, M. N. Blanco, C. V. Caceres and H. J. 8 9 K. Jiratova and M. Kraus, Appl. Catal., 1986,27, 21. J. L. G. Fierro, A. Lopez Agudo, P. Grange and B.Delmon, in 26 Thomas, Appl. Catal., 1983, 7, 139. J. H. Ashley and P. C. H. Mitchell, J. Chem. SOC. A, 1969, 2730; 10 Proc. 8th Int. Congr. Catal., Berlin, 1984, Verlag Chemie, Weiheim, 1984, vol. 2, p. 363. M. Lo Jacono and M. Schiavello, in Preparation of Catalysts, ed. B. Delmon, P. A. Jacobs and G. Poncelet, Elsevier, Amsterdam, 27 28 1968,284. J. P. Brunelle, Pure Appl. Chem., 1978,50, 1211. M. N. Blanco, C. V. Caceres, L. Osiglio, H. J. Thomas and F. J. Gil Llambias, in Actas del XVZIZ Congreso Latinoamericano de 11 12 1976, p. 473. B. R. Strohmeier and D. M. Hercules, J. Catal., 1984,86,266. A. Maezawa, Y. Okamoto and T. Imanaka, J. Chem. SOC., 29 Quimica, Santiago, Chile, 1588, vol. 2, Pontificia Universidad Catolica de Chile, p. 255. J. A. R. van Veen, P.A. J. M. Hendriks, R. R. Andrea, E. J. G. M. 13 Faraday Trans. I, 1987,83,665. H. J. Thomas, M. N. Blanco, C. V. Caceres, N. Firpo, F. J. Gil Llambias, J. L. G. Fierro and A. Lopez Agudo, J. Chem. SOC., 30 31 Romers and A. E. Wilson, J. Phys. Chem., 1990,94,5282. F. E. Massoth, Ado. Catal., 1978, 27, 265. Y. Okamoto, T. Imanaka and S. Teranishi, J. Catal., 1980, 65, 14 15 Faraday Trans., 1990,86, 2765. L. Wang and W. K. Hall, J. Catal., 1980,66, 251; 1982,77,232. S. Kasztelan, J. Grimblot, J. P. Bonnelle, E. Payen, H. Toulhoat 32 488. V. H. J. de Beer, T. H. M. van Sint Fiet, J. F. Engelen, A. C. van Haandel, M. W. J. Woles, C. H. Amberg and G. C. A. Schuit, J. 16 and Y. Jacquin, Appl. Catal., 1983,7,91. C. V. Caceres, J. L. G. Fierro, A. Lopez Agudo, M. N. Blanco 33 Catal., 1972, 27, 357. N. P. Martinez and P. C. H. Mitchell, in Proc. Climax Third Int. 17 and H. J. Thomas, J. Catal., 1985,95, 501. J. A. R. van Veen, H. De Wit, C. A. Emeis and P. A. J. M. Con$ on Chemistry and Uses of Molybdenum, ed. H. F. Barry and P. C. H. Mitchell, Climax Molybdenum Company, Ann 18 19 Hendriks, J. Catal., 1987, 107, 579. N. P. Luthra and W. C. Cheng, J. Catal., 1987,107,154. J. A. R. van Veen, P. A. J. M. Hendriks, E. J. G. M. Romers and R. R. Andrea, J. Phys. Chem., 1990,94,5275. 34 Arbor, MI, 1979, p. 105. S. Mendioroz, J. M. Palacios, J. L. G. Fierro and A. Lopez Agudo, Bull. Soc. Chim. Belg., 1987, 96,891. 20 F. M. Mulcahy, M. J. Fay, A. Proctor, M. Houalla and D. M. Hercules, J. Catal., 1990, 124, 231. Paper 4/00451E; Received 25th January, 1994