POROUS CATALYST-SUPPORT SHAPED BODY

Abstract

A porous shaped catalyst support body comprising at least 85% by weight of alpha-alumina, wherein the support has a total pore volume in the range from 0.5 to 2.0 mL/g as determined by mercury porosimetry, and a pore structure characterized by a geometric tortuosity τ in the range from 1.0 to 2.0; and an effective diffusion parameter η in the range from 0.060 to 1.0; wherein geometric tortuosity τ and effective diffusion parameter η are determined by image analysis algorithms from computer-assisted 3D reconstructions of focused ion beam scanning electron microscope analyses. The structure of the support has a high total pore volume, such that impregnation with a large amount of silver is possible, while the surface area is kept sufficiently high in order to assure optimal dispersion of the catalytically active species, especially metal species. The support has a pore structure that leads to a maximum rate of mass transfer within the support. The invention also relates to a shaped catalyst body for preparation of ethylene oxide by gas phase oxidation of ethylene, comprising at least 15% by weight of silver, based on the total weight of the catalyst, deposited on a porous shaped catalyst support body as described above. The invention further relates to a process for producing the shaped catalyst body, in which a) a porous shaped catalyst support body as described above is impregnated with a silver impregnation solution, preferably under reduced pressure; and the impregnated porous shaped catalyst support body is optionally subjected to drying; and b) the impregnated porous shaped catalyst support body is subjected to a heat treatment; wherein steps a) and b) are optionally repeated. The invention also relates to a process for preparing ethylene oxide by gas phase oxidation of ethylene, comprising the reaction of ethylene and oxygen in the presence of a shaped catalyst body according to claim 11.

Claims

1-13. (canceled)

14. A porous shaped catalyst support body comprising at least 85% by weight of alpha-alumina, wherein the support has a total pore volume in the range from 0.5 to 2.0 mL/g as determined by mercury porosimetry, and a pore structure characterized by - a geometric tortuosity τ in the range from 1.0 to 2.0; and - an effective diffusion parameter η in the range from 0.060 to 1.0; wherein geometric tortuosity τ and effective diffusion parameter η are determined by image analysis algorithms from computer-assisted 3D reconstructions of focused ion beam scanning electron microscope analyses.

15. The porous shaped catalyst support body according to claim 14, having a density in the packed tube of more than 450 g/L.

16. The porous shaped catalyst support body according to claim 14, having a BET surface area in the range from 0.5 to 5.0 m.sup.2/g.

17. The porous shaped catalyst support body according to claim 14, obtained by i) providing a precursor material comprising, based on the content of inorganic solids, - at least 50% by weight of a transition alumina having a loose bulk density of not more than 600 g/L, a pore volume of at least 0.6 mL/g and a median pore diameter value of at least 15 nm; and - not more than 30% by weight of an alumina hydrate; ii) shaping the precursor material to shaped bodies; and iii) calcining the shaped bodies in order to obtain the porous shaped catalyst support body.

18. The porous shaped catalyst support body according to claim 17, wherein the transition alumina has a loose bulk density in the range from 50 to 600 g/L and a pore volume of 0.6 to 2.0 mL/g.

19. The porous shaped catalyst support body according to claim 17, wherein the transition alumina comprises a phase selected from gamma-alumina, delta-alumina and theta-alumina.

20. The porous shaped catalyst support body according to claim 17, wherein the alumina hydrate comprises boehmite and/or pseudoboehmite.

21. The porous shaped catalyst support body according to claim 17, wherein the precursor material also comprises pore-forming materials, lubricants, organic binders and/or inorganic binders.

22. The porous shaped catalyst support body according to claim 17, wherein the precursor material is shaped to shaped bodies by extrusion, tableting, pelletization, casting, forming or microextrusion.

23. The porous shaped catalyst support body according to claim 17, wherein the calcining is performed at a temperature of at least 1100° C.

24. A shaped catalyst body for preparation of ethylene oxide by gas phase oxidation of ethylene, comprising at least 15% by weight of silver, based on the total weight of the catalyst, deposited on a porous shaped catalyst support body according to claim 14.

25. A process for producing a shaped catalyst body according to claim 24, in which a) the porous shaped catalyst support body is impregnated with a silver impregnation solution; and the impregnated porous shaped catalyst support body is optionally subjected to drying; and b) the impregnated porous shaped catalyst support body is subjected to a heat treatment; wherein steps a) and b) are optionally repeated.

26. A process for preparing ethylene oxide by gas phase oxidation of ethylene, comprising the reaction of ethylene and oxygen in the presence of a shaped catalyst body according to claim 24.

Description

[0232] The invention is illustrated in detail by the appended drawings and the examples that follow.

[0233] FIG. 1 is an illustrative surface-rendered three-dimensional FIB-SEM tomogram of a 10 .Math.m x 10 .Math.m x 10 .Math.m cubic volume of a porous metal-on-Al.sub.2O.sub.3 catalyst obtained after segmentation. The solid material marked as (i) corresponds to the Al.sub.2O.sub.3 backbone that bounds the pores that appear marked as (ii).

[0234] FIG. 2 is an illustrative surface-rendered three-dimensional FIB-SEM tomogram of a 10 .Math.m x 10 .Math.m x 10 .Math.m cubic volume of a porous metal-on-Al.sub.2O.sub.3 catalyst obtained after segmentation. The figure shows solely voxels corresponding to the supported metal.

[0235] FIG. 3 is an illustrative graphic diagram of a pore network model that has been calculated for a FIB-SEM tomogram of a 10 .Math.m x 10 .Math.m x 10 .Math.m cubic volume of a porous supported metal-on-Al.sub.2O.sub.3 catalyst. The figure shows pore regions as spheres having different volume, connected by cylindrical necks.

[0236] FIG. 4 shows the shape of the porous alpha-alumina catalyst supports A and B.

[0237] FIG. 5 shows logarithmic differential intrusion [mL/g] and cumulative intrusion [mL/g] relative to the pore size (pore diameter) [.Math.m] of the inventive porous catalyst support A.

[0238] FIG. 6 shows logarithmic differential intrusion [mL/g] and cumulative intrusion [mL/g] relative to the pore size (pore diameter) [.Math.m] of porous catalyst support B (comparative example).

[0239] FIG. 7 shows a representative scanning electron micrograph of a cross section that has been exposed by machining of the comparative catalyst 2 by means of a focused ion beam. The pores appear black, the Al.sub.2O.sub.3 backbone appears light gray, and the supported silver particles appear as white bodies. The scale bar has a horizontal length of 2000 nanometers.

[0240] FIG. 8 shows a representative scanning electron micrograph of a cross section that has been exposed by machining of the inventive catalyst 1 by means of a focused ion beam. The pores appear black, the Al2O3 backbone appears light gray, and supported silver particles appear as white objects. The scale bar has a horizontal length of 2000 nanometers.

METHOD 1: NITROGEN SORPTION

[0241] Nitrogen sorption measurements were conducted by means of a Micrometrics ASAP 2420. Nitrogen porosity was determined in accordance with DIN 66134. Before the measurement, the sample was degassed under reduced pressure at 200° C. for 16 h.

METHOD 2: MERCURY POROSIMETRY

[0242] Mercury porosimetry on the alpha-alumina shaped catalyst support bodies was conducted by means of an AutoPore V 9600 mercury porosimeter from Micrometrics (contact angle 140 degrees, Hg surface tension 485 dyn/cm, maximum head pressure 61000 psia). Mercury porosity was determined in accordance with DIN 66133.

[0243] The samples were dried at 110° C. for 2 h and degassed under reduced pressure before the analysis, in order to remove physically adsorbed species, for example moisture, from the sample surface.

METHOD 3: LOOSE BULK DENSITY OF POWDERS

[0244] In order to determine loose bulk density, the transition alumina or the alumina hydrate was introduced into a measuring cylinder via a funnel, ensuring that the measuring cylinder was not moved or agitated. The volume and weight of the transition alumina or of the alumina hydrate were determined. Loose bulk density was determined by dividing the volume in milliliters by the weight in grams.

METHOD 4: BET SURFACE AREA

[0245] BET surface area was determined to DIN ISO 9277 by means of nitrogen physisorption at 77 K. The surface area was ascertained from a 5-point BET graph. Before the measurement, the sample was degassed under reduced pressure at 200° C. for 16 h. In the case of the shaped alpha-alumina support bodies, more than 4 g of the sample was used on account of the relatively low BET surface area.

METHOD 5: DENSITY OF A POROUS SHAPED CATALYST SUPPORT BODY IN THE PACKED TUBE

[0246] Density in the packed tube was determined by introducing an amount of x g of shaped support bodies into a cylindrical glass tube having an internal diameter of 39 mm up to a mark indicating an internal tube volume of y mL. The glass tube was placed on a balance, and the increase in weight as a result of the support introduced was determined as x. The density in g/L was calculated as (x/y) x 1000.

METHOD 6: ANALYSIS OF THE TOTAL AMOUNT OF CA, MG, SI, FE, K AND NA CONTENTS IN ALPHA-ALUMINA SUPPORTS

6A. Sample Preparation for the Measurement of Ca, Mg, Si and Fe

[0247] About 100 to 200 mg (with an error of ±0.1 mg) of a support sample was weighed into a platinum crucible. 1.0 g of lithium metaborate (LiBO.sub.2) was added. The mixture was melted in an automatic melting device with a temperature ramp up to max. 1150° C.

[0248] After cooling, the melt was dissolved by cautious heating in deionized water. Subsequently, 10 mL of semiconcentrated hydrochloric acid (concentrated HCl diluted with deionized water, volume ratio 1:1, corresponding to about 6 M) was added. Finally, the solution was made up to a volume of 100 mL with deionized water.

6B. Measurement of Ca, Mg, Si and Fe

[0249] The contents of Ca, Mg, Si and Fe were determined from the solution described in point 6A by means of inductively coupled plasma optical emission spectroscopy (ICP-OES) by means of a Varian Vista Pro ICP-OES.

TABLE-US-00001 Parameters Wavelengths [nm]: Ca 317.933 Mg 285.213 Si 251.611 Fe 238.204 Integration time: 10 s Atomizer: conical 3 mL Atomizer pressure: 270 kPa Pump rate: 30 rpm Calibration: external (matrix-adapted standards)

6C. Sample Preparation for the Measurement of K and Na

[0250] About 100 to 200 mg (with an error of ±0.1 mg) of a support sample was weighed into a platinum dish. 10 mL of a mixture of aqueous concentrated H.sub.2SO.sub.4 (95% to 98%) and deionized water (volume ratio 1:4) and 10 mL of aqueous hydrofluoric acid (40%) were added. The platinum dish was placed onto a sandbath and evaporated to dryness. After the platinum dish had been cooled, the residue was dissolved by cautious heating in deionized water. Subsequently, 5 mL of semiconcentrated hydrochloric acid (concentrated HCl diluted with deionized water, volume ratio 1:1, corresponding to about 6 M) was added. Finally, the solution was made up to a volume of 50 mL with deionized water.

6D. Measurement of K and Na

[0251] The amounts of K and Na were determined from the solution described in point 6C by flame atomic absorption spectroscopy (F-AAS) by means of a Shimadzu AA-7000 F-AAS.

TABLE-US-00002 Parameters: Wavelengths [nm]: K 766.5 Na 589.0 Gas: air/acetylene Gap width: 0.7 nm (K) / 0.2 nm (Na) Atomizer pressure: 270 kPa Calibration: external (matrix-adapted standards) Method 7: Tomographic focused ion beam scanning electron microscopy (FIB-SEM) and tomogram analysis

[0252] Focused ion beam scanning electron microscopy (FIB-SEM) was used to determine porosity and the pore topology properties of the materials. Experimentally, trilobal or tetralobal cylinder bodies of the materials were cut into small aliquots (about 2 mm) with a razor blade. The smaller material fragments were infiltrated and embedded with a low-viscosity epoxy resin (Spurr, Merck) using a pyramid-tip ultramicrotomy mold, and the resin was cured at 343 K for 12 hours. The resin-embedded sample block was then cut to size and polished in an ultramicrotome (Reichert Ultracut) with a diamond blade (Diatome) and positioned on an SEM probe with an epoxy adhesive.

[0253] Next, an electrically conductive layer was applied to the resin-embedded sample and the metal stump, using a colloidal graphite suspension in isopropanol. The sample mounted on the probe was then coated by sputtering with an about 20 nm-thick carbon layer in a BAL-TEC SCD 005 coater in order to achieve fully conductive connections and to minimize local charging artefacts during SEM imaging. FIB-SEM experiments were conducted in a Zeiss Auriga dual-beam microscope equipped with a Ga ion cannon and secondary electron and backscattering electron detectors.

[0254] For minimization of curtaining effects, with the aid of the ion cannon and metal gas injection devices of the dual-beam microscope, an additional layer of metallic Pt (about 30 nm) was applied to the region of interest (ROl). Anterior and lateral channels (depth about 45 .Math.m) were machined with the focused beam of the Ga.sup.+ ions around the ROl in order to expose a finely polished anterior area (width about 30-40 .Math.m x height 35-45 .Math.m) of the block to be imaged. A cross-shaped fiducial marker was scratched on the upper surface of the sample alongside one of the lateral channels in order to serve as reference for the automatic drift correction during the FIB-SEM imaging. Next, an automatic slice & view algorithm was started in order to machine thin slices having a nominal thickness of 35-45 nm from the front face of the imaging block with the FIB cannon that was operated with an intensity of 2 nA and to record an SEM microphotograph of every newly exposed cross section with a secondary electron detector, while the electron cannon was operated at an acceleration voltage of 1-3 kV.

[0255] The collection of SEM microphotographs (2048 x 1536 pixels) was then processed by a vertical dilation correction for compensation of the angular alignment of the machining and imaging cannons in a dual-beam microscope (which form an angle of 54 degrees), and a bandpass FFT filter (see, for example, Kim, D., et al. (2019), Microscopy and Microanalysis, 25(5), 1139-1154) implemented in the FlJl-lmageJ 1.53 software was employed in order to reduce curtaining artefacts. The stack of microphotographs was aligned by means of a cross-correlation algorithm (see, for example, Yaniv Z. (2008) Rigid Registration. In: Peters T., Cleary K. (eds), Image-Guided Interventions. Springer, Boston, MA.). Finally, the stack was cut to a cubic data volume. The resulting reconstructed cubic tomograms had a volume of (20 to 23 .Math.m).sup.3 with elemental voxels of dimensions (15 to 45 nm).sup.3.

[0256] For quantification of the material topology, voxels in the reconstructed FIB-SEM tomograms were classified in accordance with their contrast (grayscale) and assigned to various subvolumes or phases, i.e. pores, alumina and supported metal, by segmentation by means of a marker-based 3D watershed algorithm (E. Dougherty, editor, Mathematical morphology in image processing, chapter 12, pages 433-481. Marcel Dekker, 1993), implemented in Avizo® (ThermoScientific), followed by a fine adjustment of the automatically recognized volumes via controlled erosion expansion functions in order to remove artefact material “islands”, and a manual threshold adjustment in order to correct local grayscale gradients that result either from curtaining effects or shadowing phenomena.

[0257] The total porosity was then determined as the proportion of the total voxels corresponding to the subvolume pores. The average geometric pore tortuosity was determined as the ratio between geodesic and euclidean pore distances by propagating a centroid path algorithm through the pores of the tomograms of the subvolume (Gostovic, D. et al., Journal of the American Ceramic Society (2011) 94: 620-627). Four quadrants of equal size and the entire volume were considered separately for two independent tomograms per sample, in order to assess statistical uncertainty associated with the average geometric pore tortuosity.

[0258] For the analysis of pore constriction, an algorithm that combines a chamfer distance transformation, a 3D watershed operation and a numerical reconstruction, as implemented in Avizo® 2020.1-XPore (ThermoScientific) (E. Bretagne (2018) Mineralogical Limitations for X-Ray Tomography of Crystalline Cumulate Rocks, Durham University), was applied to the subvolume pores in order to produce a label field that separates individual pores (local broader sections of the 3D subvolume pores connected via narrower segments or necks). The algorithm was adjusted such that voxels considered to be connected are those that have at least one common vertex, and the contrast factor marker of the H maxima was set to 2. As a result, a pore network model was constructed, i.e. a 3-D model of spherical pores that are connected by cylindrical necks that correspond to the real pore space network. The constriction parameter was defined as the square of the ratio between the average diameter for all necks and the average diameter for all pores in the pore network model.

[0259] For silver-laden catalysts, porosity, pore tortuosity, pore constriction and the effective diffusion parameters for the alumina support were quantified after the tomogram voxels that had been segmented as supported metal were first assigned to the collection of voxels corresponding to the subvolume pores, as a result of which the supported metal was mathematically removed from the surface area of the alumina support material.

Production of Porous Alpha-alumina Catalyst Supports

[0260] The properties of the transition aluminas and alumina hydrates that were used for production of porous alpha-alumina catalyst supports are listed in table 1. The transition aluminas and alumina hydrates were sourced from Sasol (Puralox® and Rural®).

[0261] Transition aluminas and alumina hydrate according to table 1 were mixed in order to obtain a powder mixture. Processing aids (Vaseline® from Unilever and glycerol from Sigma-Aldrich) and water were added to the powder mixture. Vivapur® MCC Spheres 200 (microcrystalline cellulose, from JRS Pharma) were added to the mixture. Subsequently, further water was added in order to obtain a formable precursor material. The total amounts of all components are listed in table 2.

[0262] The formable precursor material was mixed to homogeneity by means of a roller-based mixer (Mix-Muller) and then extruded with a ram extruder to give shaped bodies. The shaped bodies had the shape of a trilobe with four passages, as shown in FIG. 4. The extrudates were dried at 110° C. overnight (about 16 h) and then subjected to heat treatment in a muffled furnace at 600° C. for 2 h and then at high temperature (1475° C. for support A, 1430° C. for support B) for 4 h. The heat treatment was effected under air.

[0263] The dimensions of the dried applied layers were ascertained with a caliper gauge. The diameter of the circumscribed circle of the cross section at right angles to the support height was 11.6 cm. The term “circumscribed circle” relates to the smallest circle that fully encloses the trilobal cross section. The diameter of the inscribed circle of the cross section at right angles to the support height was 5.3 cm. The term “inscribed circle” relates to the largest possible circle that can be drawn within the trilobal cross section. The central passage had a diameter of 1.92 cm. The three outer passages had a diameter of 1.46 cm.

[0264] The resultant shaped support bodies A and B had an alpha-alumina content of more than 98% by weight, and Na, K, Mg, Ca contents below 100 ppm. The Fe content in both supports was 200 ppm. The Si content in support A was 100 ppm. The Si content in support B was 200 ppm.

[0265] The two shaped support bodies A and B had a density in the packed tube of 550 g/L.

[0266] Table 3 shows the physical properties of the supports produced according to table 2. FIGS. 5 and 6 show logarithmic differential intrusion and cumulative intrusion relative to pore size (pore diameter) for the supports produced according to table 2.

TABLE-US-00003 Transition alumina * Bulk density [g/L] Pore volume [mL/g] ** Median pore diameter value [nm] ** Puralox SCFa 140 650 0.57 10.0 Puralox TM 100/150 UF 150 0.88 18.4 Puralox TH 200/70 300 1.23 37.4 Alumina hydrates * Bulk density [g/L] Pore volume [mL/g] ** Median pore diameter value [nm] ** Pural SB1 680 0.55 8.4 Pural TH 200 340 1.20 37.6 * Puralox products are transition aluminas produced from Pural products, i.e. boehmite ** determined by nitrogen sorption

TABLE-US-00004 Support Transition alumina Alumina hydrate Pore-forming materials Processing aid Liquid A Puralox TH 200/70 173 g Puralox TM 100/150 UF 93 g Pural TH 200 67 g Vivapur MCC Spheres 200 250 g Vaseline 8.5 g Glycerol 8.4 g Water 454 g B * Puralox SCFa 140 173 g Puralox TM 100/150 UF 93 g Pural SB1 67 g Vivapur MCC Spheres 200 250 g Vaseline 8.3 g Glycerol 8.3 g Water 479 g * comparative example

TABLE-US-00005 Support BET surface area [m.sup.2/g] Pore volume [mL/g] Pore volume present in pores [mL/g] ** (proportion of total pore volume) < 0.1 .Math.m 0.1-1 .Math.m 1-10 .Math.m 10-100 .Math.m > 100 .Math.m r.sub.pv *** A 2.0 0.57 0 (0%) 0.40 (70.2%) 0.12 (21.1%) 0.04 (7.0%) 0.01 (1.8%) 0.30 B * 2.0 0.53 0 (0%) 0.22 (41.5%) 0.21 (39.6%) 0.09% (17.0%) 0.01 (1.9%) 0.95 * comparative example ** determined by mercury porosimetry *** r.sub.pv = ratio of pore volume present in pores having a diameter in the range from more than 1 to 10 .Math.m to pore volume present in pores having a diameter in the range from 0.1 to 1 .Math.m

Production of Catalysts

[0267] Shaped catalyst bodies were produced by impregnating supports A and B with a silver impregnation solution. The catalyst compositions are shown in table 4 below. The silver contents are reported in percent, based on the total weight of the catalyst. The amounts of promoter are reported in parts per million (ppm), based on the total weight of the catalyst.

[0268] Table 4: Catalyst composition (Ag contents are reported in percent by weight of the overall catalyst; amounts of promoter are reported in ppm, based on the weight of the overall catalyst)

TABLE-US-00006 Catalyst Support Ag.sub.CAT .sup.∗ [% by wt.] Li.sub.CAT [ppm] S.sub.CAT [ppm] W.sub.CAT [ppm] CS.sub.CAT [ppm] Re.sub.CAT [ppm] K.sub.ADD ** [ppm] 1 A 27.7 450 35 615 1025 1270 85 2 *** B *** 27.7 450 35 615 1025 1270 85 * the amount of silver was calculated ** K.sub.ADD means the amount of potassium applied during the impregnation, and does not comprise the amount of potassium present in the alumina support before the impregnation *** comparative example

1. Production of a Silver Complex Solution

[0269] A silver complex solution was produced according to preparation example 1 of WO 2019/154863 A1. The silver complex solution had a density of 1.529 g/mL, a silver content of 29.3% by weight and a potassium content of 90 ppm.

2. Production of Catalyst Intermediates

[0270] In each case 100.0 g of support A (intermediate 1.1) or 100.4 g of support B (intermediate 1.2) was introduced into a 2 L glass flask. The flask was connected to a rotary evaporator that was put under a vacuum pressure of 80 mbar. The rotary evaporator was set to rotate at 30 rpm. 76.55 g (intermediate 1.1) or 76.86 g (intermediate 1.2) of the silver complex solution produced in step 2.1 was added to support A (intermediate 1.1) or support B (intermediate 1.2) under a vacuum pressure of 80 mbar over the course of 15 min. After the silver complex solution had been added, rotary evaporation was continued under reduced pressure for a further 15 min. The impregnated support was then left in the apparatus at room temperature (about 25° C.) and atmospheric pressure for 1 h and mixed cautiously every 15 min.

[0271] The impregnated material was placed onto a mesh in 1 to 2 layers. The mesh was exposed to an air stream of 23 m.sup.3 (STP)/h, the gas stream having been preheated to a temperature of 305° C. The impregnated material was heated up to a temperature of 290° C. at a heating rate of about 30 K/min and then kept at 290° C. for 8 min in order to obtain Ag-containing intermediates according to table 5. The temperatures were measured by mounting three thermocouples at a distance of 1 mm below the mesh. Subsequently, the catalysts were cooled down to ambient temperature by removing the catalyst intermediates from the mesh with an industrial vacuum cleaner.

[0272] Table 5: Ag-containing catalyst intermediates (Ag contents are reported in percent by weight of the overall catalyst; amounts of promoter are reported in ppmw, based on the weight of the overall catalyst intermediate)

TABLE-US-00007 Intermediate Support Ag.sub.CAT * [% by wt.] K.sub.ADD ** [ppm] 1.1 A 18.3 56 1.2 B 18.3 56 * the amount of silver was calculated ** K.sub.ADD means the amount of potassium added during the impregnation, and does not comprise the amount of potassium present in the alumina support before the impregnation

3. Production of Catalysts

[0273] 120.5 g of the Ag-containing intermediate 1.1 or 122.2 g of the Ag-containing intermediate 1.2 as produced in step 2.2 was introduced in each case into a 2 L glass flask. The flask was connected to a rotary evaporator that was put under a vacuum pressure of 80 mbar. The rotary evaporator was set to rotate at 30 rpm. For catalyst 1, 53.80 g of the silver complex solution produced in step 2.1 was mixed with 2.16 g of promoter solution l, 2.80 g of promoter solution ll and 4.69 g of promoter solution III. For catalyst 2, 54.56 g of the silver complex solution produced in step 2.1 was mixed with 2.19 g of promoter solution l, 2.84 g of promoter solution ll and 4.76 g of promoter solution III.

[0274] Promoter solution I was obtained by dissolving lithium nitrate (Merck, 99.995%) and ammonium sulfate (Merck, 99.4%) in deionized water in order to achieve an Li content of 2.85% by weight and an S content of 0.22% by weight. Promoter solution II was obtained by dissolving tungstic acid (HC Starck, 99.99%) in deionized water and cesium hydroxide in water (HC Starck, 50.42%) in order to achieve a target Cs content of 5.0% by weight and a W content of 3.0% by weight. Promoter solution III was obtained by dissolving ammonium perrhenate (Buss & Buss Spezialmetalle GmbH, 99.9%) in deionized water in order to achieve an Re content of 3.7% by weight.

[0275] The combined impregnation solution of silver complex solution and promoter solutions I, II and III was stirred for 5 min. The combined impregnation solution was added to each of the silver-containing intermediates 1.1 or 1.2 under a vacuum pressure of 80 mbar over the course of 15 min. After the combined impregnation solution had been added, rotary evaporation was continued under reduced pressure for a further 15 min. The impregnated support was then left in the apparatus at room temperature (about 25° C.) and atmospheric pressure for 1 h and mixed cautiously every 15 min.

[0276] The impregnated material was placed onto a grid in 1 to 2 layers. A nitrogen stream of 23 m.sup.3 (STP)/h (oxygen content: < 20 ppm) was passed through the grid, the gas stream having been preheated to a temperature of 305° C. The impregnated materials were heated up to a temperature of 290° C. at a heating rate of about 30 K/min and then kept at 290° C. for 7 min in order to obtain catalysts according to table 4. The temperatures were measured by mounting three thermocouples at a distance of 1 mm below the grid. Subsequently, the catalysts were cooled down to ambient temperature by removing the catalyst bodies from the mesh with an industrial vacuum cleaner.

[0277] FIGS. 7 and 8 show representative scanning electron micrographs of a cross section that had been exposed by machining of catalysts 1 and 2 by means of a focused ion beam.

Catalyst Tests

[0278] An epoxidation reaction was conducted in a stainless steel test reactor in a vertical arrangement, having an internal diameter of 6 mm and a length of 2.2 m. The reactor was heated with hot oil that was present at a particular temperature in a heating mantle. All subsequent temperatures are based on the temperature of the hot oil. The reactor was charged with 9 g of inert steatite spheres (0.8 to 1.1 mm), onto which was packed 26.4 g of comminuted catalyst that had been sieved to a desired particle size of 1.0 to 1.6 mm, and onto that was packed a further 29 g of inert steatite spheres (0.8 to 1.1 mm). The inlet gas was introduced into the upper part of the reactor in a “once-through” mode of operation.

[0279] The catalysts were introduced into the reactor at a reactor temperature of 90° C. at a nitrogen flow rate of 130 L (STP)/h at a pressure of 1.5 bar absolute. Then the reactor temperature was increased to 210° C. at a heating rate of 50 K/h and the catalysts were kept in that state for 15 h. Then the nitrogen stream was replaced by a stream of 114 L (STP)/h of methane and 1.5 L (STP)/h of CO.sub.2. The reactor pressure was set to 16 bar absolute. Then 30.4 L (STP)/h of ethylene and 0.8 L (STP)/h of a mixture of 500 ppm of ethylene chloride in methane were added. Subsequently, oxygen was fed in stepwise until a final flow rate of 6.1 L (STP)/h has been attained. At this juncture, the inlet composition consisted of 20% by volume of ethylene, 4% by volume of oxygen, 1% by volume of carbon dioxide and ethylene chloride (EC) moderation of 2.5 ppmv (parts per million, based on volume), using methane as balance at a total gas throughput of 152.8 L (STP)/h.

[0280] The reactor temperature was increased to 225° C. at a heating rate of 5 K/h and then to 240° C. at a heating rate of 2.5 K/h. The catalysts were kept under these conditions for 135 hours. Then the EC concentration was reduced to 2.2 ppmv and the temperature was lowered to 225° C. Then the inlet gas composition was altered stepwise to 35% by volume of ethylene, 7% by volume of oxygen, 1% by volume of carbon dioxide with methane as balance, and a total gas throughput of 147.9 L (STP)/h. The temperature was adjusted so as to attain an ethylene oxide (EO) concentration in the outlet gas of 3.05%. The EC concentration was adjusted to optimize the selectivity. The results of the catalyst tests are summarized in table 6.

TABLE-US-00008 Cat. Support Time on stream *** [h] EO selectivity [%] Reactor temperature [°C] # Geometric tortuosity τ Effective diffusion parameter η ** 1 A 1.20 ± 0.02 0.075 ± 0.007 600 89.0 235 2 * B * 3.92 ± 0.16 0.006 ± 0.001 600 87.9 234 * comparative example ** effective diffusion parameter η, defined by equation 3 *** time on stream begins from the juncture of feeding of oxygen into the ethylene-containing feed stream

[0281] The error intervals correspond to the standard error SE, defined as SE = s/(N).sup.½, where s is the standard deviation and N is the number of evaluations, in the case of parameter η propagated from the uncertainties of porosity, constriction and geometric tortuosity.

[0282] It is apparent that the catalyst 1 obtained from inventive support A has much higher selectivity than the catalyst 2 obtained from comparative support B.