Removal of carbon monoxide is required in applications where the carbon oxides can deactivate catalysts such as in the production of ammonia or in low temperature fuel cells. In the case of the polymer fuel cell (PEM-FC) carbon monoxide concentrations have to be removed to less than 100ppm after the shift reaction to avoid poisoning the anode of the PEM-FC.

Conventionally, carbon monoxide and steam can, with suitable catalysts, be converted to carbon dioxide and hydrogen via the water-gas shift reaction: CO + H20 = CO2 + H 2 SH<O (generates heat) This reaction is reversible, maximum conversion being limited by the chemical equilibrium. The equilibrium is independent of pressure and the carbon monoxide concentration increases with temperature, favouring high conversions at low temperatures.

There are two types of shift reactor: high temperature and low temperature shift reactors. A typical high temperature shift (HTS) reactor operates at around 400°C. The rate of reaction is high but conversion is limited by the thermodynamic equilibrium.

The rate of conversion increases with temperature but decreases when the gas composition is close to equilibrium. Typical carbon monoxide levels that can be achieved at the exit with a HTS reactor in an ammonia plant are 2-4%. HTS reactors use Fe/Cr/A1203 catalysts, which are cheap, stable and can withstand impurities, but need temperatures of greater than 350°C to be active.

The. low temperature shift (LTS) reactors operate at the lowest possible inlet temperature to achieve the maximum carbon monoxide conversion based on the chemical equilibrium. In practice, the lowest possible inlet temperature is dictated by the dew-point (about 200°C in ammonia plant). Conventional iron based catalysts do not have sufficient activity for such low temperature operation and copper based catalysts are used. These Cu/Zn/A1203 catalysts have good activity even below 200 °C but are relatively expensive, are susceptible to poisons and have a limited temperature range. Conversion is restricted to limit temperature rise and catalyst cost.

In conventional adiabatic shift reactors, conversion in a single bed of catalyst is thermodynamically limited. As the shift reaction proceeds, the liberation of thermal energy from the forward reaction typically causes an increase in the temperature of the shift catalyst by 50-100°C. This can be higher in the case of partial oxidation process gases where the carbon monoxide content is greater. This temperature rise reduces the conversion of carbon monoxide and raises the temperature above the working range of the catalyst. To overcome this large temperature rise, the catalyst could be divided into three beds with inter-cooling or a second shift reactor could be used. The gas at the exit of the first catalyst bed will effectively reach equilibrium whenever throughput is low. The inlet and exit temperature in each bed must be inside the temperature working range of the catalysts.

To achieve low concentrations (<1 %) of carbon monoxide in the resultant gas stream, two stages of shift reaction are normally required. A high temperature shift followed by a low temperature shift.

In this two-stage shift arrangement it is necessary to lower the temperature of the process gas at the exit of the HTS reactor where it generally exceeds 400°C to a temperature of 200°C, which is suitable for the inlet to LTS reactor. Inter-stage cooling of the gas stream between reactors is achieved by heat exchange. In some cases, the temperature may be reduced, by injecting steam or condensate into the process gas. Addition of quench water allows CO concentration to be lowered but problems can arise during start up and shut down. In such plants the life of the LTS catalyst may be shortened because of the damage from the entrained water droplets and because of the presence of catalyst poisons in the water itself. Excessive condensation of water on LTS catalyst is invariably detrimental causing catalyst fragmentation and generally must be avoided. High steam levels in process gas can be tolerated provided condensation does not take place.

The catalysts used for shift reactors are generally used in pellet form. For commercially sized pellets operating at typical plant pressures, a pore diffusion limitation becomes increasingly significant at temperatures of 350°C and above.

When a reaction is highly pore-diffusion limited the effectiveness becomes inversely proportional to the pellet radius. There is a key issue of poisoning of LTS catalysts, which are sensitive to even very low levels of poisons by halides particularly as zinc and copper halides have low melting points and high mobilities. If the temperature of the catalyst is allowed to rise especially above its normal operating range sintering of the copper particles can reduce the activity of the catalyst. To allow for the deactivation of the LTS catalyst an extra 70% of the required amount of LTS catalyst is generally added to the reactor.

Conventional shift reactor designs as disclosed, for example, in Catalyst Handbook by M. V. Twigg, 2nd Edition, 1989, Wolfe Publishing, Page 292, utilise a reaction chamber packed with solid catalyst particles to perform the shift reaction. Such reactors are generally large and heavy because of their need to house solid catalyst particles and because of their brazed and welded construction. The reactors are particularly unsuitable for use in mobile applications such as the purification of reformate gas to provide hydrogen to one or more fuel cells in a vehicle.

It is an object of the present invention to provide a method of performing a shift reaction and a shift reactor which are more compact than conventional methods and shift reactors. According to a first aspect of the present invention, there is provided a method of performing a shift reaction using a diffusion bonded heat exchanger having two chambers or sets of channels formed therein, with the diffusion bonded heat exchanger being arranged to transfer heat between the two chambers or sets of channels, the first chamber or set of channels conveying the fluids performing the shift reaction and the other second chamber or set of channels conveying coolant to cool the fluid in the first set of channels.

Such a method of performing a shift reaction using a diffusion bonded heat exchanger provides for very efficient cooling of the shift reactants such that a small reactor is required.

Catalyst is preferably provided in the first chamber or set of channels. The catalyst may be provided as a layer on the chamber or set of channels. The rate of flow of coolant through the second chamber or set of channels may be controlled to ensure satisfactory cooling of the shift reactants.

According to a second aspect of the present invention, there is provided an apparatus for performing a shift reaction, the apparatus comprising a diffusion bonded heat exchanger having two chambers or sets of channels formed therein, with the diffusion bonded heat exchanger being arranged to transfer heat between the two chambers or sets of channels, the first chamber or set of channels being arranged to convey fluids performing the shift reaction, and the other chamber or set of channels being arranged to convey coolant to cool the fluids in the first set of channels.

The various aspects of the present invention may be embodied in many ways, but some specific embodiments will now be described by way of example with reference to the accompanying drawings, in which:- Figure 1 shows a diffusion bonded heat exchanger suitable for performing the first aspect of the present invention; Figure 2 shows the shift reaction being performed adjacent to the catalyst layer of a diffusion bonded heat exchanger; and Figure 3 illustrates the temperature profile and carbon monoxide concentration across a reactor of the second aspect of the present invention and a conventional shift reactor.

Description Figures 1 and 2 shows a water gas shift reactor balanced with a coolant stream in a heat exchanger. The reactor consists of a diffusion bonded heat exchanger 10 as shown in Figure 1 with two chambers or sets of channels therein, separated by a diffusion bonded heat exchanger plate. In one chamber or set of channels 11, the shift reaction takes place, in the other chamber or set of channels 12 a coolant gas stream flows as illustrated in Figure 2. A diffusion bonded heat exchanger is very compact and provides very good heat transfer between the two chambers or sets of channels.

The catalyst is in the form of a thin layer comprising catalytically active particles dispersed in an inactive matrix. The catalyst consists of copper and zinc supported on alumina. The catalyst layer is applied to the shift reactor side of the heat exchanger plate using wash-coat technology. This catalyst arrangement allows good heat transfer from the shift reaction through the heat exchanger plate to the coolant stream.

The coolant stream could be any suitable liquid or gas stream such as another gas stream in a fuel cell arrangement which requires pre-heating. The rate of flow of coolant is controlled to ensure an appropriate amount of cooling so that the shift reaction is performed under suitable temperature conditions. The temperature of the shift reaction chamber or set of channels is measured and the rate of flow of coolant is set appropriately to ensure that the shift reaction is performed at a suitable temperature. The temperature of the chamber or set of channels in which shift reaction takes place may be monitored and the rate of flow of coolant controlled appropriately. The coolant flow rate is determined dependent upon the shift reaction temperature by a processor such as a microprocessor using a look-up table or suitable algorithm. The flow of coolant may be produced by any suitable means such as a blower or pump.

Removal of the heat from the shift reactor allows the reactor to be maintained at a low temperature so that the reaction is not at equilibrium ensuring a higher CO conversion.

If the temperature is controlled to be sufficiently low as in the example of the present invention, only one shift reactor is required unlike convential shift reactions which generally require two reactors. This simplifies the plant, significantly reducing size and costs.

This method of heat removal is preferable to the traditional method of reducing the temperature in the shift reactor by using quench water. Alternatively higher steam concentrations can be used to achieve higher conversion of CO. However, the disadvantage of this is that the system is more complicated, it can lead to problems at start up and shut down and it is more important that condensation does not take place as excessive condensation of water on the LTS is detrimental causing fragmentation.

The majority of the catalyst in conventional shift catalyst particles is unused. This is because finite diffusion of reactant gases means that the reaction is complete long before gases have diffused to the centre of the particles. The smaller diffusion paths associated with thin layers means that higher reaction rates are achievable. This is manifested as a higher apparent catalyst activity and increased effectiveness factor. This is illustrated in Table 1. The reaction becomes increasingly diffusion limited for LTS pelletted catalyst as the operating temperature is increased as shown by the decrease in the effectiveness factor. For the HTS pelleted catalyst, the reaction is diffusion limited over the operating temperature of the catalyst. By using a thin layer of up to 300pm, the LTS or HTS shift catalyst has an effectiveness factor of 100% over the same temperature range. Using a thin layer reduces the amount of catalyst required to achieve a given CO conversion. Effectiveness Factor of catalyst (%) Temperature (C) 300pm thin layer LTS 5.4mm pellet HTS 8. 7mm pellet 200 100 52. 7- 250 100 33. 6 300 100 22.1 37.3 350 100 15. 3 17. 3 400 100 9. 76 4501005. 93 Table l. Superior effectiveness factor of thin layer compared to typical pelletted form of shift catalyst Increasing the effectiveness factor, makes the catalyst more resistant to poisoning since the poisoning reactions are more strongly diffusion limited than the shift reaction. By having a thin catalyst a higher amount of poison will be required to produce a given decrease in activity. By operating the shift reactor at a lower temperature, deactivation of the catalyst should be reduced as the copper based species are prone to sintering, which increases with temperature. 70% of extra catalyst is used in conventional reactors to allow for deactivation.

An increase in the catalyst activity and resistance to deactivation will significantly reduce the cost and volume of the expensive copper based catalyst required. A reduction in volume of catalyst allows the reactor to be more compact.

The pressure drop through a bed of catalyst is determined by the bed geometry and voidage. Design catalyst volumes decrease as the pressure is increased, being approximately inversely proportional to the square root of pressure for a pore diffusion limited reaction. In a traditional shift reactor the pressure is typically 30bar.

If thin film catalyst is used in the reactor the pressure drop will be very limited and lower operating pressures can be used. For example in the case of the polymer fuel cell system the pressure is 3bar. Operating at lower pressure means less expensive reactor construction materials can be used.

Example of application to solid polymer fuel cell A major technical hurdle to commercialisation of solid polymer fuel cells (PEM) is the need for a low cost and compact fuel processing system to convert hydrocarbon fuel into a hydrogen-rich gas. The inability of the PEM to tolerate more than 20ppm of carbon monoxide means there is a need for purification of the gases. The resulting gas purification plant required for a PEM fuel cell system is complicated, large and expensive. Typically the process will include a reforming or partial oxidation stage followed by high temperature shift and low temperature shift then preferential oxidation (PROx) reactor. The requirement for the PROx reaction, is a carbon monoxide concentration of <1 %. The PROx will selectively oxidise the carbon monoxide rather than hydrogen to carbon dioxide and provide the hydrogen rich fuel to the fuel cell.

A critical component of the fuel processor is the water-gas shift reactor. Existing water gas shift reactors are cumbersome because of their large size and weight. This is particularly important for fuel cells for mobile applications. A shift reactor as proposed in this invention will simplify the balance of plant reducing the cost and size.

In order to illustrate the benefit of the current invention, it is convenient to consider the output from a partial oxidation reactor, since this will contain large amounts of carbon monoxide. Likely compositions experienced by the shift reactors under typical PEM-FC conditions with 3bara system pressure are given in Table 2. HTS inlet HTS outlet LTS inlet LTS outlet Stream temp (°C) 400 479 200 255 Stream total flow (mol/s) 2.9 2.9 2.9 2.9 Mole fractions CH4 0.0003 0.0003 0.0003 0.0003 H2O 0.2006 0.1314 0.1314 0.0875 CO2 0.0330 0.1023 0.1023 0.1462 N2 0.3003 0.3003 0.3003 0.3003 H2 0.3434 0.4127 0.4127 0.4566 CO 0.1223 0.0530 0.0530 0.0091 O2 0.0000 0.0000 0.0000 0.0000 Table 2 Example of gas compositions in the shift reactors under typical PEM-FC conditions with 3bara system pressure Conventionally, a single shift reactor would achieve only a 2% concentration of carbon monoxide and overheat the catalyst to 325°C. Consequently, two reactors are needed. The table shows that such a two reactor arrangement achieved the target <1 % carbon monoxide concentration with both a HTS reactor and a LTS reactor using conventional inlet temperatures. However, figure 3 illustrates the benefit of the current invention compared to a two stage combination of an HTS reactor and an LTS reactor with beds of catalyst pellets. The compact shift reactor approach achieves the same final temperature and carbon monoxide concentration in one reactor if the partial oxidation reactor reactant pre-heat and LTS stage are integrated into a compact heat exchanger. Furthermore, the release of intrinsic kinetics in thin layer catalysts allows the use of smaller reactors. If units are sized for a 200kWe PEM-FC, the conventional packed bed approach is calculated to total 199 litres for the HTS, LTS and inter-stage heat exchanger (cooling with partial oxidation reactor reactants) without allowing for deactivation. On the other hand, integration within a compact heat exchanger yields a reactor with total volume of 84 litres, a reduction by a factor of 2.4.

In this reactor design, removal of the heat from the reactor would ensure a minimal temperature rise across the reactor. In addition, as the effectiveness of the catalyst can be improved as described in the PEM example the removal of CO to leave less than 1% concentration remaining could be achieved in one shift reactor. This will simplify and/or shrink the balance of plant required to purify a gas stream from a partial oxidation, autothermal reforming or steam-reforming reactor.