WO2001083365A2

WO2001083365A2 - Method for removal of carbon monoxide from reformate

Info

Publication number: WO2001083365A2
Application number: PCT/US2001/014224
Authority: WO
Inventors: Di-Jia Liu; Daniel R. Winstead
Original assignee: Honeywell International Inc.
Priority date: 2000-05-02
Filing date: 2001-05-02
Publication date: 2001-11-08
Also published as: AU2001257503A1; WO2001083365A3

Abstract

A processor (29) for removing carbon monoxide from a hydrocarbon reformate (28) comprises a first bed including a first bifunctional material having a first active component that preferentially adsorbs the carbon monoxide from the reformate. The first bed receives either the reformate or a reformate-oxidizing gas mixture, with the mixture being formed from the reformate and an oxidizing gas. A second bed is capable of being positionally changed relative to the first bed, with the second bed having a second bifunctional material having a second active component that preferentially adsorbs the carbon monoxide from the reformate. The second bed receives either the reformate or the reformate-oxidizing gas mixture.

Description

A CARBON MONOXIDE REMOVAL DEVICE FOR

PROTON EXCHANGE MEMBRANE FUEL CELL APPLICATION

USING A HYBRID ADSORPTION AND SELECTIVE

CATALYTIC OXIDATION APPROACH

This invention was made with Government support under Contract no. DE-FC02-99EE50578 awarded by the Department of Energy. The Government has certain rights in this invention.

CROSS REFERENCE TO RELATED APPLICATION

This application is based upon U.S. Provisional Patent Application nos. 60/100,966, filed September 18, 1998. This application is a continuation-in- part of U.S. Patent Application no. 09/399,210, filed September 17, 1999.

BACKGROUND OF THE INVENTION

. The present invention relates to an apparatus and method for removing carbon monoxide from a hydrocarbon reformate used in proton exchange membrane (PEM) fuel cells and, more particularly, to a PEM fuel cell stack. More specifically, the invention discloses an apparatus and method that provides multiple functions of capability to remove carbon monoxide through adsorption/catalytic oxidation and selective oxidation, and the combination of both, from a reformate used as fuel for PEM fuel cell application.

Energy efficient PEM fuel cell stacks use pure hydrogen for fuel and oxygen from air for an oxidant. Pure hydrogen, however, has traditionally been difficult to handle and relatively expensive to store and distribute. Consequently, the current practice is to use hydrogen rich gas mixtures obtained from reforming of various hydrocarbon fuels. For example, to obtain a convenient and safe source of hydrogen for the fuel cells, on-board reforming of hydrocarbon based fuels, such as gasoline and methanol, is expected to be utilized for automotive application. These reformed fuels, or reformate, usually contain hydrogen, nitrogen, carbon dioxide, and low levels of carbon monoxide in the range from 100's of ppm to a few percent. While the presence of carbon dioxide generally has little effect on the operation efficiency of a fuel cell, even relatively low concentrations of carbon monoxide can degrade fuel cell performance. The degradation results from the carbon monoxide chemically adsorbing over the active sites in the electrode of the fuel cell. Thus, the removal of carbon monoxide from fuel has become a major concern in the advancement of PEM fuel cell technology.

Attempts at carbon monoxide removal have been disclosed in numerous patents. One method for removing carbon monoxide from gas mixtures is known as the pressure swing method disclosed in Nishida et al., U.S. Patent no. 4,743,276 and Golden et al., U.S. Patent no. 5,531 ,809. Because the adsorption of carbon monoxide is pressure sensitive in many substances, by alternately raising and lowering the pressure within the adsorption chamber, carbon monoxide can be alternately adsorbed and purged from the adsorbent substance. There are, however, multiple drawbacks to the pressure swing absorption method, including the need for bulky and expensive pressure resistant tanks, as well as pressure and/or vacuum pump apparatus which are necessary to cycle the pressure. The parasitic weight and volume of these devices make it difficult to apply the pressure swing adsorption method for transportation applications such as a fuel cell power plant for an automobile. A second disadvantage of this approach is the significant power expenditure necessary to cycle the pressurization and depressurization steps. This additional power consumption will result in the reduction of overall efficiency of the fuel cell system. Yet another disadvantage of this process is that the toxic carbon monoxide released from desorption has to be converted to carbon dioxide with additional process steps and equipment.

Another prior art process for removing carbon monoxide involves membrane separation, whereby the hydrogen in the reformate can be separated by a metallic membrane. For example, R. E. Buxbaum, U.S. Patent no. 5,215,729 discloses a palladium based metallic membrane which provides the selectivity for hydrogen separation up to 100%. Therefore, it could remove carbon monoxide and other components from hydrogen which is the fuel for the PEFC. Although highly selective, the process has several disadvantages. Since it uses precious metal as membrane material, it is expensive. Furthermore, the reformate has to be pressurized to facilitate the separation process which results in parasitic power loss and equipment complexity. Yet another method has been referred to as preferential carbon monoxide oxidation (PROX), as described for example in Vanderborgh et. al, in U. S. Patent no. 5,271 ,916; Pow et al., U.S. Patent no. 5,316,747; and Aoyama, U.S. Patent no. 5,843,195. This process involves the preferential oxidation of carbon monoxide in the presence of hydrogen by injecting air or an oxygen containing gas into a reformate stream. The chemical reaction over the catalyst surface combines carbon monoxide and oxygen to form carbon dioxide, thereby eliminating the carbon monoxide present in the reformate. A significant limitation of the PROX method disclosed in the above patents, however, is the parasitic consumption of hydrogen. When an oxidant is introduced into a stream of reformate in a PROX device, the oxidant concentration has to be stoichiometrically higher than the amount of CO in order to reduce the carbon monoxide concentration to the acceptable level. This is particularly necessary for the automotive application where CO concentration varies significantly as the result of constant change of the load demand. The excess oxidant reacts with hydrogen; therefore, it is impossible to avoid parasitic consumption of hydrogen in the process. This problem is addressed to some degree by Buswell et al., U.S. Patent nos. 5,518,705 and 5,750,076, and Meltser et al, U.S. Patent no. 5,637,415 whereby the flow rate of oxidant being introduced into a reformate stream is limited through a regulated flow path and control scheme. By limiting the amount of oxygen introduced, the parasitic consumption of hydrogen can be reduced.

In addition to the problem of the parasitic consumption of hydrogen by the oxidizing agent, another potential shortcoming of the PROX is known as the "reverse shift" arising from the presence of carbon dioxide. This occurs when carbon dioxide reacts with hydrogen to form carbon monoxide and water. Because the "reverse shift" is temperature dependent, it is difficult to reduce carbon monoxide levels below a certain level when the temperature rise from the oxidation reaction exceeds an upper threshold. Trocciola et al., U.S. Patent no. 5,330,727, disclose a two step process to avoid the reverse shift effect. The first step involves mixing an oxidant (typically air) with the fuel, and heating it to about 320 degrees F and exposing the fuel/air mixture to a catalytic material such as dispersed platinum catalyst. The high temperature is required since the efficient oxidation of carbon monoxide cannot take place at lower temperatures in the presence of high levels of carbon monoxide. Once the carbon monoxide levels are reduced to about 60 ppm, the second step involves the more efficient oxidation at lower temperatures, where the reverse shift does not occur. Vanderborgh et al., U.S. Patent no. 5,271 ,916 disclose a similar multi-step process. Although temperature control may reduce or eliminate the "reverse shift" effect, it does not resolve the problem of parasitic consumption of hydrogen when an oxidant is mixed with a reformate to oxidize carbon monoxide. Furthermore, another significant limitation of PROX is its insufficient tolerance towards the variation of CO input level in the reformate. To minimize the parasitic hydrogen loss, the oxygen to CO ratio has to be kept at a relatively low level. Yet, the CO input level often varies as the result of change of the fuel cell power output, thus the reformate throughput. It is difficult to constantly match the CO input level with the oxygen level in a dynamic environment. Consequently, unreacted CO will leak through PROX reactor, impacting fuel cell performance downstream.

Bellows et al., U.S. Patent no. 5,604,047 disclose a process whereby a stream of hydrocarbon reformate is channeled across a bed containing an adsorbent that preferentially adsorbs carbon monoxide. Before the adsorbent has become saturated with carbon monoxide, a regeneration phase begins. Regeneration involves removing the adsorbed CO from the adsorbent preferably using a non-oxidizing sweep gas, although an oxidizing sweeping gas can be used. In either event, no discussion is apparently provided about whether the sweep gas should or must be pre-heated before it enters the bed and/or how pre-heating can effect performance. In the regeneration phase, the bed is isolated from the fuel supply, thereby avoiding parasitic consumption of hydrogen. With the preferred sweep gas being steam that is channeled through the bed, carbon monoxide can be subsequently removed through desorption. The desorbed carbon monoxide in the steam can eitherbe further oxidized through an additional step or be recycled back to a fuel processor. After regeneration of the absorbent, the sweep gas is turned off and fuel is again channeled across the bed.

Disadvantages, however, remain in this process. With the preferred use of steam, oxidation of carbon monoxide outside of the bed requires additional processing and apparatus. Steam regeneration requires the temperature of the steam substantially higher than that of the reformate which is generally already nearly or fully saturated with water vapor. Cooling after the steam regeneration can result in the condensation of water over the adsorbent which subsequently reduces or deactivates the CO removal capacity of the adsorbent. For certain materials, this type of deactivation can be irreversible due to the change of chemical property by the reaction with steam. Also, two beds alternately remove carbon monoxide from reformate and regenerate. Because both beds actively transition at the same time, there is a potential momentary lapse in fuel flow. Additionally, one bed must be "off line" (i.e., not removing carbon monoxide) for the entire time the other bed is "on line." This results in an inefficiency wherein a regenerated bed is unavailable for carbon monoxide removal until the other bed is taken off line. Furthermore, using steam to sweep CO from the adsorbent requires a relatively high amount of energy due to a large latent heat of water vaporization. Therefore, the process will reduce the overall efficiency.

Other disclosures related to carbon monoxide removal are found in Frost et al., U.S. Patent no. 5,871 ,860; Guro et al., U.S. Patent no. 5,073,356; Golden et al., U.S. Patent no. 5,126,310; and Peng, U.S. Patent no. 5,529,763.

As can be seen, there is a need for an improved method of and apparatus for removing carbon monoxide from a hydrocarbon reformate that reduces or eliminates both parasitic consumption of hydrogen and the "reverse shift" effect. Also needed is a method and apparatus that optimizes the speed, robustness and thoroughness of CO removal even when the CO level in the reformate fuel input changes significantly. An additional need is for a method and apparatus that fully utilizes a packed bed adsorbent with minimum volume and weight. Still further, a CO removal apparatus and method is needed work with a reformate stream at a temperature, pressure, and humidity similar to that used in the PEM fuel cell stack. Furthermore, the method and apparatus should be functional temporarily even when the device is substantially below the operating temperature, a critical stage commonly know as "cold start" in automotive application. Another need is for an apparatus and method that provides versatility in removing CO from a reformate such that varying applications can be addressed.

SUMMARY OF THE INVENTION

In one aspect of the present invention, a method of reducing carbon monoxide from a hydrocarbon reformate comprises mixing an oxidizing gas with the reformate to form a reformate-oxidizing gas mixture; channeling the mixture into a plurality of reactors comprising a bifunctional material, with the reactors being capable of positional change; preferentially chemisorbing the carbon monoxide by the bifunctional material; converting the chemisorbed carbon monoxide to carbon dioxide; and regenerating the chemisorption capacity of the bifunctional material in the reactors.

In another aspect of the present invention, a method of reducing carbon monoxide from a hydrocarbon reformate comprises mixing an oxidizing gas with the reformate to form a reformate-oxidizing gas mixture; channeling the mixture into at least one first reactor comprising a first bifunctional material; channeling the reformate into at least one second reactor comprising a second bifunctional material; preferentially chemisorbing the carbon monoxide by the second bifunctional material such that a chemisorption capacity of the second bifunctional material is reduced; converting the chemisorbed carbon monoxide to carbon dioxide; and regenerating the chemisorption capacity of the second bifunctional material in the at least one second reactor.

In yet a further aspect of the present invention, a processor for removing carbon monoxide from a hydrocarbon reformate comprises a first bed including a first bifunctional material having a first active component that preferentially adsorbs the carbon monoxide from the reformate, with the first bed for receiving either the reformate or a reformate-oxidizing gas mixture, with the mixture being formed from the reformate and an oxidizing gas; and a second bed capable of being positionally changed relative to the first bed, with the second bed including a second bifunctional material having a second active component that preferentially adsorbs the carbon monoxide from the reformate, the second bed for receiving either the reformate or the reformate- oxidizing gas mixture.

These and other features, aspects and advantages of the present invention will become better understood with reference to the following drawings, description and claims.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a schematic diagram of a method and apparatus for removing carbon monoxide from a fuel according to an embodiment of the present invention; Figure 2 is a schematic diagram of a multiple reactor, CO removal processor according to an embodiment of the present invention;

Figure 3a is a plan view of a CO removal processor with a rotating bed configuration according to an embodiment of the present invention;

Figure 3b is a side view of the CO removal processor depicted in Figure 3a;

Figure 3c is a perspective view of the CO removal processor depicted in Figure 3a;

Figure 4 is a side schematic view of a plug-flow reactor that can be used in the CO removal processor of the present invention; Figure 5 depicts a bifunctional material in a CO removal reactor according to an embodiment of the present invention;

Figure 6 is a graph of CO concentration versus time according to the present invention during a multiple adsorption/regeneration mode of operation; Figure 7 is a graph of CO concentration versus reactor temperature according to the present invention during a selective oxidation mode of operation.

Figure 8 is a graph of CO concentration versus time according to the present invention during an adsorption/regeneration mode of operation with sudden change of CO input concentration.

DETAILED DESCRIPTION OF THE INVENTION

Figure 1 is a block diagram of a method and system 10 for the application of removing carbon monoxide from a reformate fuel used for a PEM fuel cell stack according to one embodiment of the present invention. A hydrocarbon fuel 26 is processed by a fuel processor 27 into a hydrocarbon reformate 28 comprising hydrogen needed for fuel cell operation, as well as carbon dioxide, water, nitrogen, carbon monoxide and other trace components. The hydrocarbon reformate 28 enters the carbon monoxide removal processor 29 where carbon monoxide is extracted from the reformate 28. A substantially carbon monoxide-free reformate 32 from the removal processor 29 is then channeled to a fuel cell or stack 33 where it is used in the production of a D.C. power 34. The fuel processor 27, the removal processor 29, and the fuel cell stack 33 can be controlled by a central management system 11.

More specifically, the hydrocarbon fuel 26, such as gasoline, natural gas or methanol, is introduced into the fuel processor 27. While various designs can be used, the fuel processor 27 can preferably be constructed with a design that incorporates steam reforming, partial oxidation, or autothermal reforming stage, high temperature water-gas shift reaction stage, and low temperature water-gas shift reaction stage. The fuel processor 27 converts the hydrocarbon fuel 26 into the hydrocarbon reformate 28 by the receipt of water 12, coolant 15, and/or air 13. Typically, the fuel processor 27 will perform primary carbon monoxide reduction through the water-gas shift reactions. Additionally, as a result of receiving the water 12 and thermal management, the fuel processor 27 may also condition the humidity and the temperature of the hydrocarbon fuel 27 for subsequent carbon monoxide removal and fuel cell application. Additional humidification and temperature adjustment can be performed at the exit of removal processor 29 for optimum fuel ceil operation. The coolant 15 is used to regulate the operating temperature of the processor 27. The air, water and thermal management system 11 , such as a computer interfaced with control hardware such as valves, compressors, conduits, etc., controls the flow exchange of the water 12, air 13 and coolant 15 to the fuel processor 27. Operational data 14 that relates to the operating parameters of the fuel processor 27 is exchanged between the management system 11 and the fuel processor 27. Similarly, operational data 16 that relates to the operating parameters of the removal processor 29 and operational data 19 that relates to the operating parameters of the fuel cell stack 33 is exchanged with the management system 11. In a preferred embodiment, a single air, water and thermal management system 11 controls the fuel processor 27, the carbon monoxide removal processor 29, and the fuel cell stack 33. It is, however, envisioned that multiple control systems could be used to control each of these components.

Upon discharge from the fuel processor 27, the hydrocarbon reformate 28 enters the carbon monoxide removal processor 29. The removal processor 29, according to a preferred embodiment of the present invention, comprises a plurality of beds 30a, 30b, 30c, and 30d (Figure 2). While Figure 2 depicts only four beds or plug-flow reactors, the present invention contemplates that more or less than four beds can be utilized to practice the invention. As seen in Figure 4, all beds 30a-d are preferably of a plug-flow configuration, although other configurations are contemplated. Also, while the present embodiment includes all beds having the same configuration, the beds 30a-d can each have different configurations.

The bed or reactor 30, as disclosed in Figure 4, has an overall cylindrical configuration. At one end thereof, the reactor 30 includes a reactor inlet 53. The inlet 53 receives an initial amount of reformate 28, an oxidizing gas 24 (such as air), or a mixture of reformate 28 and oxidizing gas 24, depending on which of three modes the reactor 30 is operating, as further described below. At the end of the reactor 30 that is opposite the inlet 53 an outlet 54 enables a substantially carbon monoxide free reformate 32 or a carbon dioxide-enriched sweep gas 43 to exit the reactor 30.

The beds 30a-d contain a bifunctional material 45 (Figure 5). While the preferred embodiment contemplates each of the beds 30a-d having the same bifunctional material 45, it is also within the scope of the invention that the bifunctional materials 45 in each bed 30a-d are different. In the latter situation, for example, a first bed 30a would contain a first bifunctional material and a second bed 30b would contain a second bifunctional material. In either situation, the material 45 is "bifunctional" in that it acts as an adsorbent of carbon monoxide, an oxidation catalyst, or both, depending on which of three modes the bed 30 is operating. The bifunctional material 45 is characterized by a significant difference in Gibbs energy of adsorption between carbon monoxide and hydrogen. As a result, the bifunctional material 45 preferentially chemisorbs carbon monoxide over hydrogen that is contained in the reformate 28. The term "chemisorbs" herein refers to the quasi-chemical bonding between the bifunctional material 45 and carbon monoxide. Subsequent references herein to "adsorb" and "chemisorb" are used interchangeably.

As depicted in Figure 5, the bifunctional material 45 comprises a highly dispersed active component 50 disposed on a high, specific surface-area support 44, which may, for example, be in the form of a pellet (Figure 4). The support 44 may also rest on the surface of a substrate 46, such as a monolith or ceramic foam (Figure 5). The monolith or ceramic foam 46 may be constructed of various materials such as cordierite, alumina, zirconia, and silicon carbide. The monolith may also be constructed of corrugated thin sheets of metals such as aluminum, stainless steel, and titanium.

The active component 50 preferably comprises noble metals such as platinum, palladium, ruthenium, rhodium, iridium, gold and silver. Alternatively, the active component 50 comprises transition metals including, manganese, copper, nickel, cobalt, chromium, zinc, tin and iron. Furthermore, the active component 50 may contain combinations of any of the above metals. But the active component 50 more preferably comprises a bimettalic material, one or two noble metals and/or one or two transition metals. These metals can be in a zero valance state or oxidized state. For example, the transition metal can be in the form of metal oxides. For a noble metal based bifunctional material 45, the preferred metal loading range is from about 0.2 wt.% to 15 wt.% of the combined weight of the active component 50 and the high surface-area support 44. For transition metals, the loading can range from about 0 to 100 wt.%, but preferably falls in the range of about 0.5 wt. % to 40 wt.%.

As mentioned above, the active component 50 is "highly dispersed" on the support 44. The characteristic of a high dispersion coefficient is intrinsic to the general size and configuration of a particle, not the type of metal used. It is the ratio of the number of surface atoms in a given particle divided by the total number of atoms within the particle. As can be seen from the above formula, the dispersion coefficient is a dimensionless figure. In a preferred embodiment, the dispersion coefficient is between about 15 to 100%. A low dispersion coefficient indicates that only small fraction of the metal atoms is at the surface with poor metal utilization. This usually results in a need of increased metal loading and, hence, an increase in cost. The dispersion coefficient can be improved by reducing the metal particle size to the point that a majority of the atoms are exposed at the surface of the particles.

The support 44 is preferably a high surface area, refractory metal oxide that provides a better environment for dispersing the active component 50 to highly amorphous micro-crystallites. It also provides a stable support for the component 50 during the operation of CO removal processor 29 that often requires high temperature and humidity. Examples of suitable materials for the support 44 include metal oxides such as alumina, titania, silica, zirconia and ceria. The support 44 may also be the combination of these metal oxides. For example, it may be the mixture of ceria and alumina or titania and alumina. The characteristic of a high surface area refers to an area to weight ratio and can quantitatively vary. However, in this embodiment, the surface area preferably ranges from about 20 to 300 m²/g and, more preferably, ranges from about 100 to 300 m²/g. Below about 20 m²/g, the surface area may not be sufficient to disperse the active component 50 to fine particles.

As indicated above, the beds or reactors 30a-d operate in one of three modes: 1) adsorption/catalytic regeneration mode; 2) selective oxidation mode; and 3) hybrid mode. In either mode, after exiting the removal processor 29, the carbon monoxide-free reformate 32 enters a fuel cell 33, where it is converted into electrical energy 34 (Figure 1).

In the adsorption/catalytic regeneration mode, carbon monoxide removal is accomplished in two stages - one, where the reactors 30a-d function as an adsorbent of carbon monoxide and two, where the reactors 30a-d are regenerated through the catalytic reaction. This mode may be useful when, for example, the CO concentration is at a low level of about 100 to 1000 ppm.

In further describing the adsorption/catalytic regeneration mode, and specifically during the absorption stage of a single reactor 30, an initial amount of reformate 28 passes through the catalyst layer or bifunctional material 45 and the carbon monoxide in the reformate 28 is chemisorbed. Substantially carbon monoxide-free reformate 32 (Figures 1 and 4) is then discharged through the reactor outlet 54. Over time and in the process of adsorbing carbon monoxide, the bifunctional material 45 eventually approaches a point of carbon monoxide saturation, thereby reducing or altogether eliminating its capacity to adsorb additional carbon monoxide from the reformate 28. To allow the bifunctional material 45 to maintain an effective adsorption of carbon monoxide, the carbon monoxide must, from time to time, be removed from the bifunctional material 45, thereby regenerating the adsorption capacity of the bifunctional material 45 - i.e., the catalytic regeneration stage. The catalytic regeneration stage preferably occurs prior to the time the bifunctional material 45 reaches carbon monoxide saturation and, more preferably, before there is any substantial leakage (i.e., more than about 100 ppm) of the carbon monoxide concentration at the exit of processor 29.

During the regeneration stage of the reactor 30, the flow of reformate 28 through the bifunctional material 45 is stopped. Then, the oxidizing gas 24, such as air, flows through the bifunctional material 45. Preferably, the oxidizing gas 24 is heated upstream of the reactor 30 by a heating element (not shown), such as a component of the system 10 that can include the fuel processor 27 or the fuel cell stack 33. The heating element can also be a compressor, a heater or a heat-exchanger, which is controlled by the management system 11. The heat-exchanger can be constructed according to any well-known design such as those listed in "Handbook of Heat and Mass Transfer" N. P. Cheremisinoff eds. Gulf Publishing Co. Houston, Texas, 1986. Examples include tubular or plate-fin type heat-exchangers. The oxidizing gas 24 is pre-heated to a temperature that will initiate the catalyst reaction of the oxidizing gas with the adsorbed carbon monoxide. Additional heat generated through the initial catalytic reaction will further propagate the catalytic reaction throughout the whole bed, which leads to increased speed and thoroughness of the regeneration of a bifunctional material 45. For this embodiment, the pre-heated oxidizing gas 24 has a temperature range between about 50°C to 350°C.

In an alternative embodiment, the oxidizing gas 24 is not pre-heated and the reactor 30 thus includes an integral heater or heat-exchanger 51 that is constructed according to any well-known design, such as a coiled resistant heater or a tubular heat-exchanger. In either embodiment, and during the catalytic regeneration mode, heated oxidizing gas 24 sweeps across the bifunctional material 45, oxidizing the carbon monoxide to carbon dioxide. Because carbon dioxide has a relatively low Gibbs energy of adsorption over the bifunctional material 45, it is easily released by the bifunctional material 45 and swept away by flow of pre-heated air or sweep gas 24. With the adsorbed carbon dioxide now released, the bifunctional material 45 is again able to adsorb additional carbon monoxide. Thus, the adsorption capacity of the bifunctional material 45 has been regenerated. At the same time, a carbon dioxide-enriched sweep gas is discharged from the removal processor 29. The adsorption-regeneration cycle in a single reactor 30 comprises the steps of sequentially adsorbing carbon monoxide from the hydrocarbon reformate 28 by means of the bifunctional material 45 and catalytically regenerating the bifunctional material 45 by means of the heated air 24 sweeping through the bifunctional material 45. The preferred embodiment envisions the timing of the adsorption and regeneration stages to be controlled by the management system 11. In determining the timing of the cycle, the management system 11 will evaluate data including the amount of carbon monoxide detected at the exits of the fuel processor 27 and the CO removal processor 29, as well as the flow rate, temperature, etc. Alternatively, CO and temperature sensors may be placed inside or at the exit of each of the multiple reactors 30, with the information to be used as part of the operational data 14, 16, and 19 to the management system 11.

For increased efficiency, the CO removal processor 29 includes multiple reactors 30, as shown in Figure 2. These individual reactors 30 can be connected in either a series or parallel flow configuration, depending on the application need. For example, the multiple reactors 30a-d can be flow connected in series when maximizing the utilization capacity of each reactor 30, such as the configuration shown in Figure 2. However, the reactors 30 can be flow connected in parallel when minimizing the pressure drop across the CO removal processor 29 is important. With "N" being the number of beds in the removal processor system 29, the reactors 30 can be sequentially or serially operated at 360°/N out-of-phase with each other. For example, each reactor 30 can be sequentially regenerated one at the time, following the saturation of the previous one.

As a specific example of four reactors 30 being sequentially operable, Figure 2 schematically depicts the reactors 30a-d flow connected in series. In Figure 2, and at one point in time of an adsorption-regeneration cycle, the oxidizing air 24 flows into the first reactor 30a that is in a regeneration phase. Valves 55, 56, 57 remain closed. Thus, as the carbon monoxide adsorbed on the first bifunctional material 45 is oxidized to carbon dioxide, the oxidizing air or sweep gas 24 carries the carbon dioxide out of the first reactor 30a as a carbon dioxide or sweep gas exhaust 43. Concurrently, the reformate 28 flows into the second reactor 30b which is in an adsorption phase. Accordingly, the second bifunctional material 45 in the second reactor 30b adsorbs carbon monoxide in the reformate 28, which then enters a conduit 42d. The conduit 42d enables the reformate 28 to flow from the second reactor 30b to the third reactor 30c. The third reactor 30c is also in an adsoprtion phase and so adsorbs carbon monoxide from the reformate 28 that was not absorbed by the second reactor 30b. A conduit 41 d allows the reformate 28 to flow from the third reactor 30c and into the fourth reactor 30d which is in an adsorption phase. Consequently, the fourth reactor 30d adsorbs carbon monoxide in the reformate 28 that was not absorbed by the second and third reactors 30b, c. From the fourth reactor 30d, the carbon monoxide free reformate 32 exits. In Figure 2, the dashed line with arrows depicts the reactors 30a-d being positionally changed, thereby resulting in an operating change, such as from a regeneration phase to an adsorption phase. The positional change of the reactors 30a-d is accomplished, according to one embodiment of the present invention, by an inlet valve assembly 41 and an outlet valve assembly 42, as shown in Figures 3a-c. The inlet valve assembly 41 includes a first inlet manifold 41a that is rotationally fixed about the longitudinal axis of the removal processor 29. A second inlet manifold 41 b is juxtaposed to the first inlet manifold 41a, but is rotatable about the longitudinal axis of the removal processor 29. Each of the first and second inlet manifolds 41a, b describes four apertures 41c (not shown) that allow flow communication between the manifolds 41a, b and the four reactors 30a-d. The conduit 41d flow connects the third and fourth reactors 30c, d in this embodiment.

The inlet valve assembly 41 is fixed to a first inlet 35 and a second inlet 36. For the embodiment shown in Figures 2b-d, the first inlet 35 enables the reformate 28 to flow through a control valve 37, into the first inlet 35, and enter the second reactor 30b. The second inlet 36 enables the air 24 to flow through a control valve 38, into the second inlet 36, and enter the first reactor 30a.

Similar to the inlet valve assembly 41 , the outlet valve assembly 42 includes a fixed, first outlet manifold 42a and a rotatable second outlet manifold 42b. Each manifold 42a, b describes four apertures 42c that allow flow communication between the manifolds 42a, b and the four reactors 30a- d. The conduit 42d, in this embodiment, provides flow communication between the second and third reactors 30b, c. As shown in Figures 3b and 3c, the outlet valve assembly 42 communicates with a first outlet 39 that enables the carbon monoxide free reformate 32 to exit from the fourth reactor 30d. The valve assembly 42 also communicates with a second outlet 40 that enables the sweep gas exhaust to exit from the first reactor 30a.

In the adsorption/catalytic regeneration mode, all the control valves 55- 57 are closed so that no oxidizing gas will enter the reactors 30 b-d. In such mode of operation, it can be seen that when the first reactor 30a has completed its regeneration phase, all four reactors 30a-d can be rotated by rotation of the second inlet manifold 41b and the second outlet manifold 42b. Doing so changes the position of the reactors 30a-d relative to the first inlet inlet manifold 41a and the first outlet manifold 42a. With such positional change, the second reactor 30b can, for example, take the original position of the first reactor 30a. So changed, the second reactor 30b can undergo a regeneration phase and the first reactor 30a can undergo an adsorption phase.

In the selective oxidation mode of operation, the oxidizing gas 24 is mixed with the reformate 28 at the upstream of the reactor 30 to form a reformate-oxidizing gas mixture. Carbon monoxide in the mixture is continually adsorbed by the bifunctional material 45. Concurrently, the oxidizing gas 24 in the mixture selectively, continuously and catalytically oxidizes the carbon monoxide over the surface of the bifunctional material 45. Thereby, the chemisorption capacity of the bifunctional material 45 is not being reduced and the material 45 does not need to be regenerated. Thus, in the selective oxidation mode, a single step is used to achieve adsorption and catalytic oxidation, whereas in the adsorption/catalytic regeneration mode, two steps are required in a sequential manner. The selective oxidation approach, in essence, converts the processor 29 from a multiple regenerable adsorbent bed system into a multiple stage, selective catalytic oxidation reactor system. Such change of operation mode may be useful, for example, when there is a high concentration of CO in the reformate 28, such as in the case where carbon monoxide concentration is higher than 1%.

When multiple reactors 30 are used in the selective catalytic oxidation mode, each of the reactors 30 can remain stationary or be rotated like in the adsorption/regeneration mode above. A preferred embodiment is that the multiple reactors 30 remain stationary. The oxidizing gas is injected through the gas injection/flow control valves 55-57. The injected oxidizing gas is thoroughly mixed with the reformate at a mixing zone (not shown) before entering the portion of the reactor 30 with the bifunctional material 45. The purpose of mixing is to accomplish uniform catalytic oxidation and to achieve maximum depletion of carbon monoxide with minimum consumption of hydrogen in the reformate by the oxidizing gas. Various well known means can be used to accomplish the mixing of the oxidizing gas 24 and the reformate 28, such as a Kenics static gas mixer.

During the selective oxidation mode, the amount of oxidizing gas 24 injected through the control valves 55-57 depends on the CO concentration in the reformate flow 28 detected by the carbon monoxide sensors positioned prior to each of the reactors 30 in the processor 29. Generally, the CO concentration in the reformate flow is different at each reactor 30 with the highest concentration being in the upstream reactor 30. Consequently, the amount of oxidizing gas required at each reactor 30 is different. For example, the oxidizing gas flow regulated by valve 55 may be significantly higher than that of valve 56 when the processor 29 is operated in the configuration shown in Figure 2. Meanwhile, the valve 57 may be completely closed as a result of very low to near zero concentration of CO from the exit of reactor 30c. In addition to the dependence of CO concentration, the amount of oxidizing gas required at each stage 30 of the processor 29 also depends on the selectivity of the oxidation reaction or the volume ratio of oxidizing gas over carbon monoxide. The selectivity is defined as the amount of oxidizing gas used to oxidize carbon monoxide versus the total amount of oxidizing gas added in the reformate stream. In the reformate mixture, the remaining portion of the oxidizing gas that did not react with CO will oxidize hydrogen undesirably to form water.

The selectivity relates directly with the volume ratio. For example, when oxygen is used as oxidizing gas, the selectivity would be 50% when the volume ratio of oxygen over carbon monoxide is one, based on the stoichiometric equation of the catalytic reaction of CO and O₂. The oxygen to carbon monoxide ratio may range from 4 to 0.5 during selective oxidation which corresponds to a selectivity of 12.5 to 100%. In the preferred embodiment, the oxygen to carbon monoxide ratio ranges from about 2 to 0.75 which corresponds to a selectivity of about 25% to 67%. When the ratio is higher than about two, the amount of hydrogen consumed by the undesirable oxidation will be too high which, in turn, will reduce the fuel efficiency. When the ratio is too low, the competitive reaction of hydrogen oxidation will result in a higher amount of CO unreacted at the exit of processor 29, therefore impacting the performance of the fuel cell stack.

The hybrid mode of operating the CO removal processor 29 is a combination of the adsorption/regeneration mode and the selective oxidation mode. The hybrid mode may be useful when there is a moderate CO concentration, for example, of about 1000 to 10,000 ppm. With multiple reactors 30, one or more upstream reactors are operated in the selective oxidation mode, while one or more reactors downstream of the selective oxidation mode reactors are operated in the adsorption/regeneration mode. Thereby, a substantial portion of the CO is removed in the selective oxidation mode reactors while the adsorption/regeneration mode reactors remove the remainder of the CO. Preferably, the selective oxidation mode reactors remove about 50 to 90% of the total CO in the incoming reformate 28 so that the remaining carbon monoxide can be filtered by the adsorption/regeneration process.

As an example of the hybrid mode, and in referring to Figure 2, the reactor 30a may receive the oxidizing air 24 to be in the regeneration phase of the adsorption/regeneration mode. At the same time, and for the selective oxidation mode, the control valve 55 remains partially open so that the reactor 30b receives an appropriate amount of oxidizing air 24 to mix with the reformate 28. The amount of oxidizing gas is determined by the CO concentration in the incoming reformate 28 detected by the carbon monoxide sensor and the stoichiometric ratio ranges from about 2 to 0.5 when the oxidizing gas is oxygen. Concurrently, and as part of the adsorption/regeneration mode, the valve 56 remains closed to the flow of oxidizing air 24 into the reactor 30c. However, the reactor 30c does receive the reformate 28 from the upstream reactor 30b whereby the reactor 30c is in the adsorption phase. The reactor 30d can also be in the adsorption phase of the adsorption/regeneration mode whereby the valve 57 is closed and the reactor 30d receives the reformate 28 from the upstream reactor 30c. Regeneration occurs upon the detection of the carbon monoxide concentration exceeding the designed level at the exit of processor 29. According to the example shown in Figure 2, the regeneration can be accomplished by rotating the bed position so that the freshly regenerated reactor 30a can replace 30d and maintain the high CO removal efficiency of the processor 29.

It can be seen that the present invention also provides a method for reducing carbon monoxide from a hydrocarbon reformate. In one embodiment, the oxidizing gas 24 is mixed with the reformate 28 to form a reformate-oxidizing gas mixture. The mixture is then channeled into at least one first reactor 30 comprising a first bifunctional material 45. Alternatively, the reformate 28 (in the absence of the oxidizing gas) is channeled into the one first reactor. In either event, the reformate 28 (in the absence of the oxidizing gas) is channeled into at least one second reactor 30 comprising a second bifunctional material 45.

When there is a channeling of the mixture, such channeling of the mixture and channeling of the reformate occur concurrently. The carbon monoxide is preferentially chemisorbed by the first and second bifunctional materials 45 such that a chemisorption capacity of the second bifunctional material 45 is reduced. The chemisorbed carbon monoxide is oxidatively converted to carbon dioxide in the first and second reactors. However, only the chemisorption capacity of the second bifunctional material in the reactors 30 is regenerated. In further describing the method of the present invention, the reactors are disposed in one of a series flow configuration and a parallel flow configuration. Upon the conversion to carbon dioxide, the carbon dioxide is discharged from the reactors.

EXAMPLES

Figure 6 is a graph of carbon monoxide concentration versus time according to the selective oxidation mode of the present invention. A bifunctional material with approximately 3.5 wt. % Ru over gamma alumina was prepared by the wet incipient method, which is well known in the art and described, for example, in "Heterogeneous Catalysis in Industrial Practice" by Charles N. Satterfield, McGraw-Hill, New York, 1991. About 3.2 g of the bifunctional material was packed into a bed with a plug flow reactor configuration. The packing volume was 10cc. A mixture of synthetic reformate containing 1250 ppm carbon monoxide, 20% carbon dioxide, 37% hydrogen (H₂), and the balance nitrogen (N₂) was channeled through the plug flow reactor at 130°C and a flow rate of 1 liter per minute. The effluent from the reactor was directed to carbon monoxide and carbon dioxide non- dispersive infrared (NDIR) detectors downstream from the reactor, thus the change of carbon monoxide and carbon dioxide concentrations could be monitored constantly. Multiple cycles of adsorption/regeneration were conducted.

The concentrations of carbon monoxide and carbon dioxide are disclosed in Figure 6. In each cycle, carbon monoxide in the reformate was nearly completely removed. The non-zero baseline for carbon monoxide shown in Figure 6 was due to the infrared absorption interference from excess amounts of carbon dioxide present. At about the 10% breakthrough point (which is defined as the point where the CO concentration at the exit of the reactor is 10% of that in the incoming reformate), the stream of reformate was diverted from the plug flow reactor, and hot air at 120°C was introduced. Quick carbon monoxide oxidation was observed, as indicated by the sharp peak of carbon dioxide. Consistent capacity and activity in removing and oxidizing carbon monoxide in the multiple adsorption/regeneration cycles were shown in Figure 6. .

In another example, the bifunctional material from above was packed into a reactor. A humidified synthetic reformate mixture containing 1250 ppm carbon monoxide, 17.5% carbon dioxide, 32% hydrogen (H₂), 12.7% water, and the balance nitrogen (N₂) was channeled through the plug flow reactor while the temperature was ramped up from 50 to 350°C. The reformate was directed into the reactor with a gas hourly space velocity of 14,400 hr¹. Air was mixed into the reformate at concentrations from 1 to 3%.

Near complete depletion of CO concentration was observed in the temperature range from 150 to 280°C in all three air concentrations, as shown in Figure 7. This demonstrated the capacity to reduce carbon monoxide under the selective oxidation mode.

Yet in another example, the bifunctional material from above was packed into a reactor. A humidified synthetic reformate mixture containing 1250 ppm carbon monoxide, 17.5% carbon dioxide, 32% hydrogen (H₂), 12.7% water, and the balance nitrogen (N₂) was used for the demonstration. The synthetic reformate mixture was first bypassed the reactor to register the CO concentration through CO NDIR detector. During the bypass mode, a 30 second injection of enriched CO stream into reformate mixture results in a 40% increase of CO concentration in the stream as is shown in Figure 8, simulating the change of CO concentration in reformate due to the change of fuel reformer. The same 30 second CO injection was repeated during the adsorption mode. However, no leakage was observed as the result of higher input CO concentration. This example demonstrated the excellent tolerance of the reactor towards the transient change of carbon monoxide input level.

As can be appreciated by those skilled in the art, the present invention provides an improved method of and apparatus for removing carbon monoxide from a hydrocarbon reformate and regenerating a catalytic material that reduces or eliminates both parasitic consumption of hydrogen and the "reverse shift" effect. The method and apparatus provide CO removal capability operable at the ambient temperature and tolerable toward dynamic variation of carbon monoxide input level. Also provided is a method and apparatus that optimizes the speed, robustness and thoroughness of the regeneration of CO removal capacity and optimizes the oxidation of carbon monoxide. Additionally, the present invention is an apparatus and method that provides versatility in removing CO from a reformate such that varying applications can be addressed. In particular, the present invention provides a single apparatus that allows one of three modes of carbon monoxide removal to occur. It should be understood, of course, that the foregoing relates to preferred embodiments of the invention and that modifications may be made without departing from the spirit and scope of the invention as set forth in the following claims.

Claims

WE CLAIM:

1. A method of reducing carbon monoxide from a hydrocarbon reformate, comprising: mixing an oxidizing gas with said reformate to form a reformate- oxidizing gas mixture; channeling said mixture into a plurality of reactors comprising a bifunctional material, said reactors being capable of positional change; preferentially chemisorbing said carbon monoxide by said bifunctional material; converting said chemisorbed carbon monoxide to carbon dioxide; and regenerating said chemisorption capacity of said bifunctional material in said reactors.

2. The method of Claim 1 , wherein said reactors are disposed in one of a series flow configuration and a parallel flow configuration.

3. The method of any one or more of Claims 1-2, wherein the step of channeling said mixture comprises concurrently flowing said mixture through said reactors.

4. The method of any one or more of Claims 1-3, further comprising discharging said mixture from said reactors.

5. The method of any one or more of Claims 1-4, wherein the step of discharging comprises removing said carbon dioxide from said reactors.

6. The method of any one or more of Claims 1-5, wherein the step of regenerating occurs continuously.

7. The method of any one or more of Claims 1-6, wherein said bifunctional material comprises a support characterized by a specific surface area between about 20 to 300 m²/g.

8. The method of any one or more of Claims 1-7, wherein said bifunctional material comprises a support selected from the group consisting of alumina, titania, silica, zirconia, ceria, and mixtures thereof.

9. The method of any one or more of Claims 1-8, wherein said bifunctional material comprises a substrate selected from the group consisting of a monolith and a foam substrate, both of which are constructed of a material selected from the group consisting of cordierite, alumina, zirconia, silicon carbide stainless steel, aluminum, and titanum.

10. The method of any one or more of Claims 1-9, wherein said bifunctional material comprises a noble metal present between about 0.2 to 15 wt.% of the combined weight of the noble metal and support materials.

11. The method of any one or more of Claims 1-9, wherein said bifunctional material comprises a transition metal present between about 0.5 to 40 wt.% of the combined weight of the transition metal and support materials.

12. The method of any one or more of Claims 1-9, wherein said bifunctional materal comprises a bimetallic material having a plurality of metals selected from the group consisting of noble metals and transition metals.