Conversion of natural gas to liquid hydrocarbons ("Gas To Liquids"or"GTL" process) is based on a 3 step procedure consisting of : 1) synthesis gas production; 2) synthesis gas conversion by FT synthesis; and 3) upgrading of FT products (wax and naphtha/distillates) to final products.

The Fischer-Tropsch reaction for conversion of synthesis gas, a mixture of CO and hydrogen, possibly also containing essentially inert components like C02, nitrogen and methane, is commercially operated over catalysts containing the active metals Fe or Co. Iron catalysts are best suited for synthesis gas with low H2/CO ratios (< 1.2), e. g. from coal or other heavy hydrocarbon feedstock, where this ratio is considerably lower than the consumption ratio of the FT- reaction (2.0-2. 1). The present invention is concerned with Co-based catalysts, in particular, supported Co-based catalysts. A variety of products can be made by the FT-reaction, but from supported cobalt, the primary product is long-chain hydrocarbons that can be further upgraded to products like diesel fuel and petrochemical naphtha. Byproducts can include olefins and oxygenates.

To achieve sufficient catalytic activity, it is customary to disperse the Co on a catalyst carrier, often referred to as the support material. In this way, a larger portion of Co is exposed as surface atoms where the reaction can take place.

The present invention is concerned with alumina, as a support material.

Supported cobalt catalysts are the preferred catalysts for the FT synthesis. The most important properties of a cobalt FT catalyst are the activity, the selectivity usually to Ce+ and heavier products and the resistance towards deactivation.

Known catalysts are typically based on titania, silica or alumina supports and various metals and metal oxides have been shown to be useful as promoters.

In a paper by Iglesia et al. ["Selectivity Control and Catalyst Design in the Fischer-Tropsch Synthesis: Sites, Pellets and Reactors"Advances in Catalysis, Vol 3,1993] a Thieles modulus is defined as a product of two components, #n and X, where Tn depends only on the diffusivity and reactivity of the individual molecules, whereas % depends only on the physical properties and site density of the catalyst. They have described a model whereby the selectivity to C5+ products can be described as a volcano plot in terms of X. The structural parameter is given as: % = Ro2 q) omlrp v where 6m is the site density, e. g. as the number of surface atoms of Co metal atoms per cm2 of pore area in the catalyst particle, Ro is the diffusion length, i. e. the radius of an essentially spherical catalyst particle, C is the porosity of the particle (cm3 pore volume/cm3 particle volume) and rp is the mean pore radius.

This expression suggests that % only depends on universal constant, characteristic data for cobalt in the catalyst as well as the size and density of the catalyst particles. It is particularly significant that % does not depend on the pore radius, rp. However, it now appears that the selectivity of the Fischer- Tropsch reaction to C5+ products indeed does in fact depend on the pore size.

In a paper by Saib et al. ["Silica supported cobalt Fischer-Tropsch catalysts: effect of pore diameter of support"Catalysis Today 71 (2002) 395-402], the influence of the effect of the average pore diameter of a silica support on the properties of a cobalt catalyst and their performance in F-T synthesis is discussed. The article concludes that the support pore diameter has a strong effect on cobalt crystallite size with larger crystallites forming in larger pore sizes. Also, the activity was found to be a function of the metal dispersion and the maximum Ce+ selectivity a function of the conversion.

In EP 1 129 776 Al it is argued that internal diffusion phenomena in a catalyst particle depend on the chemical and morphological structure of the catalyst (pore dimensions, surface area, density of the active sites) and on the molecular dimensions of the species in question. This is a general teaching found in relevant textbooks, e. g. expressed in terms of the Thiele modulus, and it is significant that the pore dimension, i. e. the pore radius or diameter is one of the critical parameters. Further, it is taught that for the Fischer-Tropsch synthesis, interparticle diffusion will create low concentrations of CO towards the centre of the particle with a consequent progressive rise in the H2/CO ratio inside the catalyst and that this condition favours the formation of light hydrocarbons (lower a-value and C5+ fraction). On the other hand, it is stated that multiphase reactors of the slurry type generally use small catalyst particles (20-150 mm) which do not give internal diffusion problems, and more specifically that for catalysts based on differently supported cobalt used in the Fischer-Tropsch synthesis, it is possible to neglect internal diffusion limitations by operation with particles having diameter of less than 200 mm.

In EP 0 736 326 B 1, it is shown that the C5+ selectivity can increase over a certain range of increasing pore size for a cobalt on alumina type FT catalyst.

However, no reference or details of the method of measuring pore size is given, and it is well known that reported values vary significantly with method, e. g. for different probe gases or whether adsorption or desorption isotherms are employed.

In general, after impregnation of an alumina carrier with a solution of a cobalt catalyst material, the carrier is dried and calcined at a relatively low temperature of 200 to 450°C, eg. at 300°C, for 2 to 16 hours. However, it is known that prolonged calcination at higher temperatures, eg. above 500°C can reduce catalyst activity.

This is in the first instance due to agglomeration of Co crystallites giving a reduced Co surface area for the FT-reaction, but at higher temperatures Co reacts with the alumina itself to form an inactive spinel phase, cobalt aluminate CoAl204. Transformation to cobalt aluminate was demonstrated by Davis and co-workers (Applied Catalysis, Volume 247, Pages 335-343,2003) to occur at 650°C and to transform completely at 850°C. The present invention relates to the surprising beneficial effect of high temperature treatment of an impregnated catalyst carrier on its attrition level.

According to one aspect of the invention, there is provided a method of producing an alumina-supported catalyst, which comprises the following steps: a first impregnation step in which an initial alumina support material is impregnated with a source of a 2-valent metal capable of forming a spinel compound with alumina ; a first calcination step in which the impregnated alumina support material is calcined at a temperature of at least 550°C to produce a modified alumina support material; a second impregnation step in which the modified alumina support material is impregnated with a source of catalytically active metal ; and a second calcination step in which the

impregnated modified support material is calcined at a temperature of at least 150°C.

According to another aspect of the invention, there is provided a method for the production of an alumina support for a catalyst, which comprises impregnating an initial alumina support material with a source of a 2-valent metal capable of forming a spinel compound with alumina, and calcining the impregnated alumina at a temperature of at least 550°C.

The invention also extends to catalyst material made in accordance with the first aspect and a support material made in accordance with the second aspect.

The invention also extends to the use of the support material and the use of the catalyst in an F-T synthesis.

The F-T synthesis may be conducted in a fixed bed reactor or a non-fixed bed reactor, such as a slurry bubble column.

The invention also extends to a process for the production of hydrocarbons which comprise subjecting H2 and CO gases to a Fischer-Tropsch synthesis reaction in a reactor in the presence of a catalyst of the invention. The process can be conducted under typical F-T operating conditions. The products can be subjected to various post-processing operations, such as de-waxing, hydro- isomerisation and/or hydro-cracking.

Amorphous catalyst support materials typically have specific surface areas between 50 and 500 m2/g, more typically between 100 and 300 m2/g. The starting alumina materials used in the present invention are all, at least predominantly, of the y-alumina type, preferably with specific surface areas

between 150 and 200 m2/g. These supports can be prepared by spray-drying techniques of an appropriate solution in order to obtain essentially spherical particles of appropriate size, e. g. 80 % in the range between 30-120 pm. After spray-drying, the material is calcined at a high temperature to give the appropriate crystal size and pore structure.

It is also important that the total pore volume is sufficiently high, above 0.2 cm3/g or better, above 0.4 cm3/g, or even above 0,6 cm3/g. The pore volume is often measured by the BET method applying nitrogen as the adsorption gas.

This method does not take into account large pores where a mercury porosimeter is more relevant. A less accurate, but more practical parameter is the measured water absorbtivity, which can be directly correlated with the amount of cobalt that can be impregnated on the catalyst by the incipient wetness procedure. A high pore volume will give a light material suitable for operation in a slurry environment and ease the impregnation by minimising the number of impregnation steps required. At the same time the support, and the final catalyst, should have sufficient strength for extended operation of months and years with minimal attrition of the materials. This can be tested in a slurry environment or by the ASTM method applicable for testing FCC (fluid catalytic cracking) catalysts.

The properties of y-alumina catalyst support materials have been investigated, both as received from the alumina vendor (Sasol GmbH of Hamburg, Germany), and following heat treatment. The properties are set out in Table 1.

Table 1. Properties of alumina as received and heat treated. Alumina Calcination Surface Pore Pore Attrition Water ab-a- batch temperature area volume (N2) diameter [g/50g Sorbtivity alumina no. [oC] [m2/g] [ml/g] [nm] at 5h] [ml/g]** [%] Al-1 500* 191 0,721 11,4 4,9 1,23 0 "1100 55 0, 390 22, 0 7, 0 0, 85 17 "1130 15 0, 084 20, 4 15, 2 0, 69 85 " 1150 7 0,019 11,4 39,0 0,60 Al-2 500* 170 0,734 12,9 4,8 1,30 0 " 900 107 0,684 19,1 4,9 1,15 " 1100 56 "1130 18 0, 108 22, 2 0, 75 87 "1140 10 0, 049 18, 3 26, 3 0, 71 97 " 1140 7 0,026 14,8 26,0 0,74 100 * Equivalent to"as received"material. The average particle size is respectively 62 pm and 59 pLm.

* * The water absorbtivity is determined as the"incipient wetness"point by successively adding water to the point where water is visually released from the pores of the alumina by gentle tapping of the beaker containing the material.

The as received y-alumina materials from the alumina vendor have been treated in one or both of two ways, namely, further heat treatment/calcinations at temperatures in the range 500-1200 °C ; and impregnation by one or a mixture of substances followed by heat treatment in the 700-1300°C temperature range. The properties of the resulting materials have been investigated.

Upon high temperature treatment, the y-aluminas or the different alumina hydrates will be converted to transition phase aluminas, denoted 6, % or K - aluminas, that all finally will be converted to a-alumina, with a gradual decrease in specific surface areas. These aluminas may also be suitable as support materials for cobalt for the Fischer-Tropsch synthesis, even though this may have specific surface areas in the range 10-50 m2/g. The specific surface areas and pore volumes must be balanced towards the requirements for sufficiently high cobalt metal loading and dispersion. However, it is also possible to increase the high temperature surface stability of aluminas by adding certain stabilising agents like lanthanum (lanthanum oxide). In this way, the y-phase can be retained, even above 1000 °C. Other stabilising agents have been used, such as magnesia and ceria.

Three other properties are important for the support, and for the final catalyst, in particular when used in a slurry reactor environment such as a slurry bubble column. These are the particle density, the particle size, and the abrasion or attrition resistance. The density is important in order to achieve a suitable distribution (dispersion) of the catalyst particles in the reactor. A light material is particularly advantageous for avoiding catalyst settling or an excessive concentration of particles in the lower part of the reactor. The particle size also affects settling and the catalyst concentration profile, but should not be excessively small, since this makes separation of the liquid product from the reactor slurry more difficult and increases the incidence of particles being transported with the gas phase at the top of the reactor. Attrition should be minimised to prolong the life of the catalyst and to avoid contamination of the liquid hydrocarbon product.

Examples of these parameters for selected y-alumina supports and their heat- treated analogues are given in Table 1. As can be seen, there is a clear

tendency for the attrition resistance to diminish significantly as the calcination temperature exceeds 1100°C simultaneously with an increase of the a-alumina content. It is stated in WO 02/47816 that catalysts with low surface areas and/or high a-alumina contents lead to a significant increase in the selectivity to the desired C5+ products in the FT-synthesis. As seen, however, the pore volume is reduced and attrition can reach an unacceptable level.

As mentioned above, an important method for characterisation of the present catalysts and support materials is related to the determination of the pore characteristics; pore volume, pore size distribution, surface area and average pore size. Various techniques may be applied, but most are variations of the so-called BET method using the adsorption or desorption isotherms of a gas that is adsorbed to the surface of the pores, typically nitrogen, but also certain noble gases or mercury are used for specialised purposes. In the present invention, the nitrogen desorption isotherm is used, measured by standard procedures on an ASAP 2000 instrument from Micromeretics and the standard included software for calculation of the pore characteristics.

The values reported in the Tables in this specification are the BET surface area, the BJH desorption cumulative pore volume between 17 and 3000 A diameter and the BJH desorption average pore diameter (4V/A). Specific values will vary depending on the method, but the general findings are expected to be valid independent of the method employed, and characterisation data compared therefore should be harmonised to a given method.

The modified ASTM type equipment for testing attrition consists of two main parts, one air feeding system and one reactor where the attrition takes place.

Compressed air passes through a pressure regulator (5 bar) to a moisture chamber where the air is moistened to approximately 30 % relative moisture.

This is done to avoid static electricity in the system. The amount of air is then adjusted in a mass flow controller. The humid air then enters the reactor through a sieve tray where the holes have a diameter of 0.4 mm. Because of these holes, the gas reaches sonic velocity, which causes the"wear and tear"on the particles in the reactor. The reactor has an internal diameter of 35.6 mm (1.4 inches) and a length of 711 mm (28 inches) and the pressure is approximately 1.8 bar. After passing through the reactor, the velocity is lowered in a separation chamber which has an internal diameter of 112 mm (4.4 inches) and a length of 305 mm (12 inches). There is a conical connection 203 mm long (8 inches) between the reactor and the separation chamber.

Particles > 40 urn will fall back down into the reactor, while smaller particles < 40, um (fines) will enter a Soxhlet-filter through a u-formed tubing, connected to the separation chamber via a conical connection 106 mm long (4 inches). A vibrator is mounted on the separation chamber, to loosen any particles on the inside walls.

50 g of powder or catalyst, sieved to > 40 um before testing, is loaded to the reactor, and the reactor is connected to the separation chamber. The air is turned on, and the fines produced in the reactor and collected in the Soxhlet filter are weighed every 15 minutes during the first 2 hours, and every 30 minutes during the next 3 hours. A normal run lasts 5 hours and the amount of fines produced can be plotted against time.

Support modification and catalyst preparation Table 2. Properties of alumina modified by metal salt impregnation and high- temperature treatment. Al-2 is the starting material. Impregnated Calcination Surface Pore Pore Attrition a-Water ab- metal and level temperatur area volume (N2) diameter [g/50g at alumina sorbtivity [wt%] * e [m2/g] [rnl/g] [nm] 5h]** [%]*** [oC] 5 Co 900 107 0,62 18 4,3 1,00 "1050 44 0, 30 23 4, 6 0, 74 1100 12 0, 07 23 12 0, 61 "1140 9 0, 05 22 20 10 Co 900 100 0,61 18 4,5 1, 00 " 1050 47 0,34 24 4,8 0, 70 " 1100 14 0,15 40 5,5 0, 54 "1140 10 0, 09 33 6, 9 0, 50 20 Co 300 144 0,49 11 6,8 0, 92 "900 76 0, 47 20 7, 0 0 0, 90 " 1140 10 0,09 33 3,4 12 0, 47 10 Al 900 90 0,42 14 8,8 1, 12 " 1050 55 0,34 18 5,0 0,80 "1140 6 0, 02 15 28, 0 0, 78 10 Zn 1120 17 0,16 36 10,5 54 0,57 "1140 12 0, 13 38 6, 9 63 0, 56 " 1160 11 0,11 37 6,9 64 0,53 12,5 Zn 1140 14 0, 16 39 4, 6 0, 52 15 Zn 1140 13 0,15 42 3,4 0,47 5 Mg 1140 17 0,16 32 3,0 0, 56 10 Mg 1120 36 0,21 18 0, 57 "1140 29 0, 18 19 0, 3 21 0, 51 " 1160 16 0,12 22 0. 5 37 0, 41 5Mg+5Zn 1140 15 0,14 32 0,5 0, 70 10 Mn 1140 3 0,01 12 16, 4 0, 53 3 Ni 1140 9 0, 07 28 6, 2 0, 50 S Ni 1140 13 0, 10 27 3, 6 78 0, S "1150 10 0, 09 33 0, 5 "1160 8 0, 04 20 0, 5 "1170 9 0, 06 27 0, 5 "1180 8 0, 05 21 3, 0 0, 45 5 Ni**** 1140 18 0, 15 29 2, 9 0, 50 7, 5Ni**** 1120 31 0,21 19 1,4 37 0,51 7, 5 Ni 1140 20 0, 17 28 2, 8 57 0, 51 10 Ni 900 93 0, 38 14 10 Ni 1140 18 0,165 30 1,5 0, 45

* Nominal values assuming complete reduction to metal. Note that the samples here all are in oxide form.

** Bold face signifies lower attrition than the alumina carrier calcined at the same temperature.

***Simplified estimation from peak heights of most intense peak for each compound in standard Cu-XRD. The other XRD peaks for the samples calcined at 1140°C correspond to the MeAL204 spinel.

**** On Al-1 type alumina.

From Table 2 it can be seen that a standard catalyst in calcined form, (the 20Co/300 °C sample), has an attrition level higher than the support material, that is to say, 6.8 g compared to 4.8 g. In other words, a significant reduction in strength is observed by impregnation of cobalt nitrate and standard calcination. What has now been astonishingly found is that the strength increases significantly by further calcination to 1140 °C (3.4 g), an attrition level which is actually lower than for the starting alumina material. The

comparison is even more compelling with the alumina support calcined at the same temperature (1140°C), as most of the strength then is lost (26.3 g). The positive effect on the catalyst strength increases with calcination temperature and Co loading.

For comparison, impregnation of aluminium nitrate was attempted, but it can be seen that little effect on the attrition was gained as in fact the attrition level of 28.0 g for the impregnated sample is slightly above the 26.3 g obtained for the support, both calcined at 1140 °C. On the other hand, Zn has a positive effect on attrition, very much in line with Co. It therefore seems as if impregnation with a metal salt and high temperature calcination enhances the strength of alumina if the added metal is able to form a well defined crystalline mixed oxide with aluminium, in this case a spinel with composition MeA1204, Me being a 2-valent metal. An even stronger effect is found using an Mg salt, as an extremely strong material is formed. The attrition level after 5 h in a jet cup of 0.5 % is below anything seen for a catalyst or catalyst support. The origin of the effect can only be speculative, but it would seem that the magnesium ion is smaller than Co2+ or Zn2+ and therefore may exhibit higher solid-state diffusion rates at the given temperature. This points to a possible positive effect of modifying alumina with other metal salts/oxides containing small metal ions, like lithium. Nevertheless, it was surprisingly found that nickel gives a very attrition resistant modified alumina support, approaching the values for magnesium (Figure 1).

Unless otherwise stated, all the catalysts used contain a nominal amount of cobalt of 20 wt% and 0,5 wt% Re, calculated assuming reduced catalysts with complete reduction of cobalt. The actual metal loading as determined by XRF or ICP may vary up to 10 %, i. e. for cobalt between 18 and 22 w% of the total reduced catalyst weight. Other promoters besides rhenium, specifically,

platinum, iridium or ruthenium, can be employed. It is also possible to add a second promoter such as lanthanum oxide or a mixture of oxides of the lanthanides or other compounds which are difficult to reduce.

Before impregnation, the catalyst support may be precalcined at about 500 °C.

Impregnation is usually in one step, but multiple steps can also be employed, from a mixed aqueous solution of appropriate metal salts, generally of cobalt nitrate and perrhenic acid or alternatively ammonium perrhenate. The impregnation technique generally used is the pore filling or"incipient wetness" method, in which the solution is mixed with the dry support until the pores are filled. The definition of the end point of this method may vary somewhat from laboratory to laboratory so that an impregnated catalyst could have a completely dry appearance or a sticky snow-like appearance. However, in no instances are there are any free flowing liquid present.

A number of alternative impregnation procedures are known in the art which use alternative solvents and chemicals, however, in the present invention, the preferred procedure involves aqueous incipient wetness with solutions of cobalt nitrate (Co (N03) 2 * 6 H20) and perrhenic acid (HRe04). Possible alternatives include using cobalt acetate (s), cobalt halide (s), cobalt carbonyl (s), cobalt oxalate (s), cobalt phosphate (s), organic cobalt compounds, ammonium perrhenate, rhenium halide (s), rhenium carbonyl (s), industrial metal salt solutions, organic solvents, etc. Furthermore, the impregnation technique may encompass all available methods besides incipient wetness, such as precipitation, impregnation from slurry with surplus liquid, chemical vapour deposition etc. It is well known that the impregnation method may influence the dispersion of the active metal (cobalt) and hence the catalytic activity, but as the Fischer-Tropsch reaction is believed to be non-structure sensitive, dispersion should not influence the selectivity

The impregnated catalyst is dried, typically at 80-120 °C, to remove water from the catalyst pores, and then calcined at typically 200-450°C, e. g. at 300°C for 2- 16h.

Catalyst testing One critical step before testing is the activation of the catalyst that involves reduction of cobalt oxide (s) to cobalt metal. This reduction can be performed by flowing a suitable reductive gas over the catalyst particles. Particularly suitable are hydrogen or carbon monoxide or mixtures thereof. The reductive gas can be mixed with inerts like nitrogen, noble gases or steam and suitable temperatures and pressures should be applied. If a fluidised bed reactor is used for activation, it may be convenient to use a recycle of (part of) the reductive gas and a slight atmospheric total overpressure in order to achieve a suitable gas flow. It is also possible to use elevated total pressures, eg. up to 8 bar or higher, or even the Fischer-Tropsch reactor pressure. Selection of the reduction temperature strongly depends on the actual catalyst formulation, in particular on the presence and nature of promoters. It is well known that Re is highly efficient as a promoter in achieving high reducibilities at a conveniently reduced temperature.

Some fixed-bed testing of activated catalysts has been performed in a laboratory unit with four parallel fixed-bed reactors. About 1 g of catalyst particles in a controlled size fraction was mixed with 5 times the volume of inert SiC. Reduction was performed in situ with hydrogen before an approximate 2: 1 mixture of hydrogen and CO was carefully added. After 20 h on stream at 210°C and 20 bar total pressure, the space velocity was adjusted to give an estimated conversion level of CO after 90 h of between 45 and 50 %. It is of utmost importance to perform selectivity comparisons, as well as activity

comparisons, at the same conversion level, since the level of steam generated in the reaction has a profound influence on catalyst performance.

Table 3. Catalyst performance in the Fischer-Tropsch reaction. Modified alumina Catalyst Relative Relative Attrition support Al-2 impregnation. activity C5+ resistance of [metal impr. wt% Composition selectivity support /calc. Temp. °C] [wt%/wt%] ---/500 20Co/0,5Re 1,54 0,933 Good ---/900 20Co/0,5Re 1,46 0,930 Good ---/1100 20Co/0,5Re 1,53 0,945 ---/1130 20Co/lRe 1, 02 0, 991 Low ---/1140 20Co/0, 5Re 0, 75 0, 988 Very low ---/1140 20Co/0, 25Re 0, 83 0, 972 Very low 5Co/900 20Co/0, 5Re 1, 68 0, 938 Good 5Co/900 20Co 1, 27 0, 914 Good 5Co/1050 20Co/0, 5Re 1, 56 0, 950 Good 5Co/1050 20Co 1, 10 0, 945 Good 10Co/900 20Co/0,5Re 1,50 0,941 Good 10Co/900 20Co 1, 14 0, 927 Good 1 OCo/1050 20Co/0, 5Re 1, 37 0, 961 Good lOCo/1050 20Co 0, 82 0, 956 Good 20Co/900 20Co/0,5Re 1,73 0,952 Acceptable ---/500 12Co/0, 5Re 0, 92 0, 931 Good ---/1130 12Co/lRe 1, 05 0, 987 Low ---/1140 12Co/lRe 0,66 1,007 Very low 5Co/1100 12Co/0, 5Re 0, 91 0, 985 Low lOCo/1100 12Co/0, 5Re 0, 83 0, 989 Good lOCo/1140 12Co/0, 5Re 0, 69 0, 990 Acceptable lOCo/1140 12Co 0, 54 0, 977 Acceptable 20Co/1140 12Co/0, 5Re 0, 44 0, 986 Very good 10Al/1050 20Co/0,5Re 1,37 0,953 Good 10A1/1050 20Co 0, 58 0, 946 Good 10Zn/1120 24Co/1Re 1,32 0,974 Low l OZn/900 20Co/0, 5Re 1, 48 0, 922 10Zn/1120 12Co/0,5Re 0,95 0,975 Low lOZn/1140 12Co/0, 5Re 0, 86 0, 987 Acceptable 10Zn/1140 12Co/0,25Re 0,84 0,979 Acceptable lOZn/1140 12Co 0, 68 0, 963 Acceptable 10Zn/1160 12Co/0,5Re 0,82 0,985 Acceptable 12, 5Zn/1140 12Co/0,25Re 0,67 0,977 Good 15Zn/1140 12Co/0, 5Re 0, 76 0, 984 Very good 10Mg/1140 24Co/1Re 0,64 0,932 Extremely good 5 Mg/1140 12Co/0, 5Re 0, 48 0, 979 Very good 10Mg/1120 12Co/0,5Re 0,53 0, 954 lOMg/1140 12Co/0, 5Re 0, 38 0, 952 Extremely good 10Mg/1160 12Co/0,5Re 0,25 0, 928 5Mg+5Zn/1140 12Co/0, 5Re 0, 44 0, 972 Extremely good 3Ni/1140 12Co/0, 5Re 0, 66 0, 991 Acceptable 5Ni/1140 12Co/0, 5Re 0, 89 0,989 Very good 5Ni/1150 12Co/0, 5Re 0, 87 0, 980 5Ni/1160 12Co/0, 5Re 0, 75 0, 992 5Ni/1170 12Co/0, 5Re 0, 62 0, 983 5Ni/1180 12Co/0, 5Re 0, 66 0, 992 Very good 5Ni*/1140 12Co/0, 25Re 0, 73 0, 970 Very good 7, 5Ni*/1120 12Co/0, 5Re 1, 04 0, 957 Extremely good 7, 5Ni/1140 12Co/0, 25Re 0, 74 0, 965 Very good 10Ni/900 12Co/0,5Re 1,48 0,906 lONi/1140 12Co/0, 25Re 0, 78 0, 973 Extremely good *On Al-1 type alumina.

In Table 3 the first column relates to the first impregnation and calcination, showing the impregnated metal (with"---"indicating no impregnation) and the temperature of calcination. The second column relates to the second impregnation (with cobalt) and calcination, showing the cobalt and rhenium contents. Calcination was carried out at about 300°C.

From Table 3 it is clear that heat treatment of the Al-2 support at 900 °C or above, and particularly above 1100 °C, increases the selectivity of the catalyst, as also described in the present Applicants'WO 02/47816. It would seem that the same effect is seen for all modified aluminas that have been impregnated with a metal solution and then high temperature treated before standard impregnation and calcination. It should also be noted that the addition of Rhenium gives improved activity and selectivity for all catalysts in accordance with previous reports. Thus the present invention means that it now is possible to combine high selectivity with high strength of the catalyst particles.

There are, however, some differences between the 2-valent modifiers that have been applied. As it is desirable to use as little modifier as possible, values can e. g. be compared for the same wt% of the added 2-valent modifier before calcination at 1140 oC. First it can be noted that at the 10 wt% level only magnesium and nickel is able to give an attrition classified as"good"or better, and that this classification even is maintained for a 5 wt% loading. However, magnesium gives a clearly inferior catalytic performance, whereas 5 wt% nickel shows excellent activity and selectivity.

A particular feature can be observed in the pore size distribution diagram of the catalysts (Figure 2), or the modified supports, plotted as differential pore volume as a function of pore diameter from the BET measurements with nitrogen. The pore size distribution of the Al-2 is bimodal, but the degree of bimodality decreases with calcination temperature of the modified support, concurrent with enhancement of strength and C5+ selectivity.

Figure 1 shows the effect of Ni loading and calcination temperature on the strength of modified Al-2 supports. Figure 2 shows temperature effect on the pore size distribution of catalysts employing the modified catalyst supports, prepared by high temperature calcination of 5 wt% Ni as nickel nitrate impregnated on y-alumina.