Fuel cells are being investigated in many places as a possible energy source for driving vehicles and for stationary generation of electricity. The use of fuel cells is still highly dependent on the availability of the fuel: hydrogen (H2). It is not to be expected that an infrastructure for hydrogen will be set up within the foreseeable future. Especially for mobile applications, it is therefore necessary to transport an available fuel, or a fuel that becomes available, and to convert this to hydrogen as the feed for the fuel cell.

A gas mixture that consists mainly of hydrogen and carbon dioxide (COz) is then produced-for example via steam reforming and/or partial oxidation-from fuels such as methane, LPG, methanol, petrol, diesel and other hydrocarbons. Said gas mixture, which is rich in hydrogen, is then fed to the fuel cell which generates electricity by an electrochemical reaction of hydrogen with oxygen.

However, a certain amount of carbon monoxide (CO) is also always liberated during the conversion of said fuels into hydrogen. For instance, a gas mixture of, for example, 75 % (V/V) H2,24 % (V/V) C02 and 1 % (V/V) CO is produced on steam reforming of methanol. A solid polymer fuel cell, the major candidate for transport applications, is extremely sensitive to CO, which even in low concentrations (0.01 % (V/V)) has an adverse effect on the performance of the fuel cell. For a usable system it is therefore necessary to remove CO down to the said level and preferably down to a lower level (< 0.005 % (V/V), 50 ppm). A technically attractive option for removing CO from H2-containing gas streams is by means of selective oxidation of CO to C02 at low temperature (100 °C-200 °C). In this context it is important that the consumption of hydrogen by non-selective oxidation to water is minimised.

The power of ruthenium (Ru) to catalyse the oxidation of CO is, for example, known from the ammonia synthesis process. Thus, it is known from US Patent 3216782 (9 November 1965) that 0.5 % (m/m) Ru on alumina (A1203) is capable of oxidising 0.055-0.6 % (V/V) CO in the presence of H2 at between 120 °C and 160 °C to a level of less than 15 ppm. In this case it is necessary that the quantity of oxygen (02) added is such that the molar 02/CO ratio is between 1 and 2. The excess oxygen which is not needed for the

oxidation of CO reacts with hydrogen to give water. It has not been investigated whether this Ru catalyst is also capable of oxidising CO from a typical reformate gas to a CO level of 15 ppm under the same conditions (temperature, 02/CO ratio).

In the Journal of Catalysis 142 (1993), Academic Press Inc., pages 257-259, S. H. Oh and R. M. Sinkevitch describe 0.5 % (m/m) Ru/y-A1203 as highly effective in the complete oxidation, at low temperature (100 °C), of 900 ppm CO with 800 ppm oxygen (02) in a gas mixture which also contains 0.85 % (V/V) H2, with the remainder being N2. Data on the stability of the Ru catalyst are not given in the article and in addition the behaviour of the catalyst in a realistic reformate gas containing H2, CO2, H20 and CO in much higher concentrations was not investigated.

Current state of the art European Patent EP 0 743 694 Al (20 November 1996) refers to an oxidation unit for the selective oxidation of CO in H2-rich gas at a reaction temperature of between 80 °C and 100 °C. A molar ratio of 02/CO of 3 is used. The final CO content is a few ppm. The excess oxygen reacts with hydrogen to give water. The catalyst consists of a 0.2 % (m/m)- 0.5 % (m/m) Pt-Ru alloy on A1203. No examples which would show the stability of the catalyst are given.

US Patent 5 674 460 (7 October 1997) describes a structured reactor for the catalytic removal of CO from H2-rich gas at between 90 °C and 230 °C. Depending on the temperature, the catalyst in this case consists of Pt on y-A1203, Pt on zeolite-Y or Ru on y-A1202. The invention is explained solely on the basis of 5 % (m/m) Pt on y-A1203, by means of which the CO content can be reduced to about 40 ppm at a reaction temperature of between 80 °C and 130 °C. No stability data are given in this patent either.

In the Journal of Catalysis 168 (1997), Academic Press, pages 125-127, R. M. Torres Sanchez et al. describe gold on manganese oxide as an alternative catalyst for the oxidation of CO in H2 at low temperatures (approximately 50 °C). In particular the price, due to the high gold loading (approximately 4-10 % (m/m)), makes the use of this type of catalyst less interesting. Moreover, this type of catalyst is able to withstand carbon dioxide to only a limited extent.

It is not clear from the above whether the catalysts of the prior art are suitable for the selective oxidation of CO in H2-rich reformate gas mixtures where there is high activity in conjunction with good stability in the temperature range 100 °C-200 °C and where a low

oxygen excess can be used to minimise the hydrogen consumption.

Discovery of new catalyst One aim of the present invention is to provide a method for the selective catalytic oxidation of CO from H2-rich, COz-and H20-containing (reformate) gas mixtures, making use of as small as possible an amount of oxygen and at relatively low temperature. A further aim of the present invention is to provide a catalyst which has high chemical and thermal stability and can be produced in a cost-effective manner by means of a simple method of preparation from commercially available starting materials and a low noble metal loading.

The use of commercially available a-A203 as carrier material in the preparation of 0.5 % (m/m) Ru on A1203 led, surprisingly, to a catalyst which in the temperature range 120 °C-160 °C combines high activity (> 99% conversion of CO) with high stability (a CO conversion of at least 97 % for a period of at least 50 hours) in the oxidation of CO with a relatively small excess of oxygen in dilute reformate gas. These results were found to be appreciably better than the results which were obtained with a commercially available 0.5 % (m/m) ruthenium catalyst with y-AI203 as the carrier (specific surface area > 100 m2/g), which is representative of the catalysts used in the abovementioned studies and reflects the prior art.

It has also been found that the addition of Pt and the lowering of the total noble metal loading resulted in a catalyst which showed even better stability for the selective oxidation of CO in both dilute and undiluted reformate gas (a CO conversion of at least 99 % for a period of at least 50 hours).

It has furthermore been found that in particular the nature and the specific surface area of the A1203 carrier used are the factors determining the exceptional performance of the Ru and Ru-Pt catalysts according to the present invention. Preferably, alumina is used in the form of a-Al203. A highly active and stable catalyst is formed when the specific surface area of the a-Ah03 is in the range from 3 m2/g to 25 m2/g.

The catalysts in the present invention can be prepared in a simple manner via a standard impregnation method from commercially available starting materials. Compared with the current state of the art, the method according to the present invention has the following advantages: -complete oxidation of CO to C02 in the temperature range 120 °C to 160 °C with only a small excess of oxygen (O2/CO = 1) compared with the stoichiometrically required quantity

of oxygen (02/CO= 0.5), -minimal hydrogen consumption as a result of low oxygen excess (02/CO = 1), -stable action at 130 °C in simulated reformate gas (0.5 % (V/V) CO, 0.5 % (VN) 02, 74 % (VN) H2, 19 % (V/V) C02 and 6 % (V/V) H20) for a period of at least 50 hours (residual quantity of CO < 50 ppm), -low noble metal loading of less than 0.5 % (m/m). a-Al203 is a commercial product that is used, inter alia, in the electronics industry in the production of thick and thin substrate layers by tape casting. Another application is the production of industrial ceramics.

The use of this a-Al203 as carrier for a selective oxidation catalyst for CO in H2-rich gas mixtures has not been described before.

The invention will be explained in more detail on the basis of the following examples together with the appended figures.

In the figures: Figure 1 shows the activity of a 0.5 % (m/m) Ru-on-a-A1203 catalyst (code AlRu-5) compared with the activity of a commercial Ru catalyst with 0.5 % (m/m) Ru on y-A1203 (code GlRuC-5) in the oxidation of CO in dilute reformate gas, Figure 2 shows the stability of AlRu-5 in the CO oxidation at 130 °C compared with the stability of GlRuC-5 in dilute reformate gas, Figure 3 shows the activity of a 0.25 % (m/m) Ru, 0.125 % (m/m) Pt-on-a-Al203 catalyst (code AlRuPt-48) compared with the activity of AlRu-5 in the oxidation of CO in dilute reformate gas, Figure 4 shows the stability of AlRuPt-48 in the CO oxidation at 130 °C compared with the stability of AlRu-5 in dilute reformate gas, Figure 5 shows the activity of AlRuPt-48 in the oxidation of CO as a function of the reformate gas composition, Figure 6 shows the stability of AlRuPt-48 in the CO oxidation at 130 °C as a function of the reformate gas composition and Figure 7 shows the activity of A2RuPt-48 compared with the activity of AlRu-5 and G3Ru-5 in the oxidation of CO in undiluted reformate gas.

In the following tests the Ru-on-oc-A1203 and the Ru-Pt-on-a-A1203 catalysts were prepared by impregnation of a commercial a-A1203 carrier with solutions of the salts

ruthenium nitrosylnitrate and hexachloroplatinic acid. The effect of the a-A1203 carrier on the CO oxidation activity and stability of the catalyst is determined under m below. The effect of the addition of Pt and the lowering of the total noble metal loading on the catalyst activity and stability is given under IV. Finally, the activity and the stability of the catalyst as a function of the composition of the reformate gas are determined under V.

I. Preparation of Ru-on-a-A1203 and Ru-Pt-on-a-A1203 catalysts The catalysts according to the present invention were prepared by dry impregnation of a-Al203 powder with solutions of ruthenium nitrosylnitrate ( (Ru (NO) (NO3) x (OH) y (x+y=3), Ru content of the solution 1.5 % (m/m)) and hexachloroplatinic acid (H2PtCI6. xH20), Pt content 0.5 % (m/m)).

The 0.5 % (m/m) Ru-on-a-A1203 catalyst (code AlRu-5) was prepared by adding 5 gram of the Ru solution to 15 gram of the a-Al203 powder in a glass beaker and then stirring well until a pasty substance was formed. This paste was then dried in air in an oven for 16 hours at 80 °C. During drying the setting paste was stirred several times. After drying, the solid material was finely ground to a homogeneous powder with the aid of a mortar. The powder thus produced was then pressed to give a pill. After crushing the pill in a mortar a 0.25 mm to 0.5 mm sieve fraction was prepared for the catalytic measurements. The catalyst prepared was stored in a polyethene sample bottle at room temperature.

In the case of the 0.25 % (m/m) Ru and 0.125 % (m/m) Pt-on-a-Al203 catalysts (codes AlRuPt-48 and A2RuPt-48), first 1.68 gram of the Ru solution and then 2.51 gram of the Pt solution were added to 10 gram of the a-Al203 powder. The subsequent preparation steps were identical to those described above for AlRu-5.

II. Test apparatus and test procedure The conversion of CO was studied in an automated micro-flow set-up operating under atmospheric pressure. The following gases were available to the set-up: N2,02, H2, CO2, CO and H20. It was possible to measure the gases H2, C02 and CO with the aid of a Perkin- Elmer model 8500 gas chromatograph equipped with a methanizer, connected in series, a TCD and an FID. A pneumatically controlled 6-way tap was used for sampling the product gas. CO was also measured occasionally with an Elsag Bailey Hartmann & Braun model URAS 1 OE ND-IR analyser.

The precursor was contained in a Pyrex glass reactor having an internal diameter of

10 mm. The catalyst bed was covered with glass wool and a layer of glass beads. The height of the catalyst bed was approximately 5 mm, whilst the gas flow was approximately 75 ml/min. The space velocity (SV) was approximately 11,000 h-1 in this case. The amount of precursor required (0.25 mm-0.5 mm fraction) was 200 or 400 mg. The temperature was measured immediately below the catalyst bed using a CrAl thermocouple.

During the measurements the catalyst sample was exposed to a pre-mixed gas containing 0.5 % (V/V) CO, 0.5 % (VN) O2, 5 or 19 % (V/V) C02, 15, 51 or 74 % (V/V) H2, 6 or 7 % (V/V) H20, with the remainder being N2. Prior to the CO oxidation measurement the catalyst sample was pre-treated with, successively, air at 400 °C and 25 % (V/V) H2 in N2 at 550 °C for activation. The activated catalyst was then cooled under H2/N2 to the starting temperature for the test. The reactor was flushed with N2 for approximately 10 minutes each time the gas composition was changed. For activity measurements the starting temperature was always 80 °C, after which the reactor temperature was raised in 10 °C steps to a final temperature of 250 °C. The CO conversion was determined at each temperature. For stability measurements the catalyst bed was first brought to the measurement temperature under H2/N2 after the pretreatment, after which the CO conversion was determined once an hour for a period of 50 hours. The general test conditions for the CO oxidation measurements are given in Table 1.

The CO conversion was calculated on the basis of the amount of CO in the product gas (COout) using the GC and the amount of CO in the feed gas (COin = 0.5 % (V/V)) determined using the GC in accordance with: CO conversion (in %) = 100 + (COin-CO. ut)/COin. Using the NDIR it was separately determined that the detection limit of the GC for CO was approximately 25-30 ppm.

Table 1 General test conditions Weight of catalyst sample 200-400 mg Volume of catalyst bed approx. 0.4-0.6 ml Particle size 0.25-0.50 mm Gas flow rate 75 ml/min Spatial velocity of the gas per hour (GHSV) 10,000-15,000 h- Feed gases Reformate gas 1 0.5% CO, 0.5% 02, 15% H2,5% C02, 7% H20, remainder N2 Reformate gas 2 0.5% CO, 0.5% 02, 51% H2, 5% CO2, 7% H20, remainder N2 Reformate gas 3 0.5% CO, 0.5% 02,74% H2, 19% CO2, 6% H20 Total pressure atmospheric Temperature of catalyst 80 °C-250 °C (10 °C steps in the activity measurements) bed 130 °C (stability measurements) , III. Effect of a-AI203 carrier on CO oxidation in dilute reformate gas The test results for the oxidation of CO with 02 in dilute reformate gas (gas 1) over the Ru-on-alumina catalyst (code AlRu-5) show that the use of a-Al203 as the carrier for Ru results in both a better activity and a better stability in the oxidation of CO compared with a commercial Ru-on-y-A1203 catalyst (code GlRuC-5).

Figure 1 shows the activity of AlRu-5 compared with the activity of GlRuC-5 in the oxidation of CO in dilute reformate gas 1. Catalyst AlRu-5 achieves a more complete CO conversion over a wider temperature range than does GlRuC-5.

Figure 2 shows the stability in the CO oxidation in dilute reformate gas 1 with AlRu-5 compared with the stability of GlRuC-5. AlRu-5 is found to be both more active and more stable in the CO oxidation than GlRuC-5 over a measurement period of 50 hours.

IV. CO oxidation in dilute reformate gas with Ru-Pt on a-A1203 Test results for the oxidation of CO with 02 in dilute reformate gas I over a Ru-Pt-on- a-Al203 catalyst (code AlRuPt-48) demonstrate that the addition of Pt and lowering the total noble metal loading results in a catalyst which is more stable than the AlRu-5 described above. The addition of platinum and lowering the total noble metal loading was not found to have a significant effect on the activity of the catalyst.

Figure 3 shows the activity of AlRuPt-48, which has a low loading, compared with the activity of AlRu-5 in the oxidation of CO in dilute reformate gas 1. It can clearly be seen that from 120 °C AlRuPt-48 shows virtually the same CO conversion as a function of the temperature as AlRu-5. This is despite the lower noble metal loading of AlRuPt-48 compared with AlRu-5.

Figure 4 shows the stability of AlRuPt-48 in the CO oxidation at 130 °C compared with the stability of AlRu-5 in dilute reformate gas 1. AlRuPt-48 displays a higher conversion of CO than AlRu-5 over the entire measurement period.

V. Effect of reformate gas composition on CO oxidation with Ru-Pt on a-A1203 Test results for the oxidation of CO with 02 in various reformate gases 1,2 and 3 over the Ru-Pt-on-a-A1z03 catalyst described above show that activity and stability are virtually independent of the composition of the reformate gas.

Figure 5 shows the activity of AlRuPt-48 in the oxidation of CO measured in various reformate gas compositions. Only at the highest temperatures is the conversion of CO in the less dilute reformate gases 2 and 3 somewhat lower than the conversion in the most dilute reformate gas 1.

Figure 6 shows the stability of AlRuPt-48 in the CO oxidation in the three different reformate gases 1,2 and 3. The very high CO conversion with this catalyst is dependent to only a very slight extent on the composition of the reformate gas; even with simulated undiluted reformate gas 3 there is more than 99% CO conversion over the entire measurement period (residual quantity of CO < 50 ppm).

Figure 7 shows the activity of three catalysts in the oxidation of CO in undiluted reformate gas 3. The various curves in Figure 7 for 0.5 % (m/m) Ru on y-Al203 (code G3Ru-5), 0.5 % (m/m) Ru on a-Al ffi (code AlRu-5) and 0.25 % (m/m) Ru and 0.125 % (m/m) Pt on a-Al ffi (code A2RuPt-48) show the substantial effect of the type of carrier material (y compared with a) and the metal composition (Ru compared with Ru/Pt).