WO2014108223A1

WO2014108223A1 - Method for regeneration of sulfur-poisoned fuel cell stacks

Info

Publication number: WO2014108223A1
Application number: PCT/EP2013/069475
Authority: WO
Inventors: Thomas Rostrup-Nielsen; Rahul Singh; Niels Erikstrup; Thomas Heiredal-Clausen; Jeppe Rass-Hansen; Karen WONSYLD
Original assignee: Topsøe Fuel Cell A/S
Priority date: 2013-01-11
Filing date: 2013-09-19
Publication date: 2014-07-17

Abstract

A method for regeneration of asulfur-poisoned fuel cell stack, preferably a solid oxide fuel cell (SOFC) stack, comprising applying a voltage potential to a fuel cell stack for a period of time sufficient to increase the average cell potential of the sulfur-poisoned fuel cell stack. The regeneration is initiated after a detected performance loss, and the regeneration can be a full regeneration, partial regeneration or a full regeneration and an improvement.

Description

Method for Regeneration of Sulfur-poisoned Fuel Cell Stacks

The present invention relates to a method for regeneration of a sulfur-poisoned fuel cell, fuel cell part, fuel cell stack or fuel cell stack system, especially a solid oxide fuel cell (SOFC) stack. More specifically, the invention concerns a method for regeneration of sulfur-poisoned fuel cell stacks by applying a voltage potential, current pulses or a combination of both to the stack.

The method can be used for regeneration of single cells, one or more stacks, an assembly of stacks, systems of stacks, several systems of stacks etc. When current is drawn from a solid oxide fuel cell stack, a deposition of sulfur on the surface of the nickel anode is normally observed. The binding of sulfur to the anode sur^¬ face is very strong. It only takes a minor content of e.g. ¾S in the feed gas to provoke almost 100% sulfur coverage of the surface of the nickel anode. Even with only a few ppm ¾S in the feed gas, close to 50% of the surface be^¬ comes covered. This means that if a sulfur-free gas is passed over the anode, then the adsorbed sulfur will only leave the anode surface extremely slowly.

Experiments made by the applicant have shown that there may be a certain positive effect of an electric potential on the sulfur tolerance when running a solid oxide fuel cell, because the sulfur tolerance is considerably higher for an SOFC than it is for a nickel catalyst (powdered Ni anode) with no potential applied. The higher the initial potential is, the higher sulfur tolerance can be observed. This ef- feet could also be true for the cathode side by applying an embodiment of the method of the invention the opposite way.

It is known that reaction rates or reaction paths change when a potential is applied to a catalyst surface, either by electrocatalysis or by electrochemical promotion. Elec- trocatalysis is the simplest electrochemical effect of a catalytic reaction. In this case the ions, which are trans^¬ ported to the three-phase boundary by an applied potential, participate directly in the reactions taking place there.

This effect can be described by the Nernst equation. Elec^¬ trocatalysis is what occurs in regular fuel cell or battery reactions . Regarding the electrochemical promotion or NEMCA (non-

Faradaic electrochemical modification of catalytic activi^¬ ty) , this is what occurs when the catalytic activity is changed much by an applied voltage and the formation of an electric double layer. The ions and electrons do not move across the interface (at least not because of the catalytic reaction) , which means that the reactions are purely chemi^¬ cal, but electrochemically promoted. Hence, this effect cannot be described by the Nernst equation. In fact the definition of NEMCA is that the catalytic effect is much greater than what a Faradaic contribution can do. The NEMCA effect is very much affected by both the amplitude and the direction of the potential.

Two possible reaction sequences could individually be the reason why an electric potential results in a higher sulfur tolerance when running a solid oxide fuel cell. It is also possible that they both apply. The first of these reaction sequences has to do with SO2 formation. Thus, a nickel catalyst can be regenerated by oxidation, possibly through the reactions

NiS + H₂0 NiO + H₂S (1) and

H₂S + 2 H₂0 «■ S0₂ + 3 H₂ (2)

As the equilibrium constant is low (3.5-10^-8 at 700°C), predominantly caused by reaction (1) (k_pi = 3.5-10^-7), only a small amount of hydrogen can prevent this regeneration from taking place. However, at the anode side of a fuel cell the potential causes an oxidizing environment, perhaps giving the possibility for these reactions to occur.

There are not any previous studies of this for solid oxide fuel cells. For proton exchange membrane (PEM) fuel cells with Pt catalysts, however, some studies show an effect of applying a potential on sulfur desorption with S0₂ formation .

The second of the above reaction sequences relates to the water gas shift reaction (WGS reaction) . The electrochemical promotion of the water gas shift reaction or the re- verse reaction over polarized YSZ has been studied by two groups both using Pt as catalyst, and both groups find that CO binds more strongly to Pt when 0^2~ has migrated to the three-phase boundary, whereas it binds less strongly when it has migrated away from the three-phase boundary, leaving 0^2~ vacancies. In these studied cases the result is that the rate of the WGS reaction increases when 0^2~ has migrat^¬ ed away from the three-phase boundary, thus opposed to what occurs at the anode of an SOFC. Another group (Catalysis Commun. 1_5, 6-9 (2011)) has made a study on a nickel cata^¬ lyst with a K⁺ conducted carrier and found that migration of positive charge to the three-phase boundary caused an increased rate of the WGS reaction, whereas negative vacan^¬ cies at the three-phase boundary caused a decrease for this reaction. None of these three studies included sulfur poi^¬ soning . When a fuel cell stack is operated using sulfur-containing fuel (e.g. sulfur-containing species H2S, S02, COS and or^¬ ganic sulfur-containing components) and a current is drawn from the stack, a deposition of sulfur on the surface of the anode is normally observed. This is especially true for nickel-containing anodes. Electrochimica Acta 5_5, 5683-5694 (2010) and Journal of Catalysis 199, 247-258 (2001) dis^¬ close the adsorption and desorption of sulfur species from a Pt electrode of a proton exchange membrane (PEM) fuel cell and the effect of electrochemical cleaning.

Fu et al . , Journal of Power Sources 187, 32-38 (2009), have investigated the potential dependence of sulfur dioxide poisoning and oxidation at the cathode of PEM fuel cells by cyclic voltammetry (CV) and electrochemical impedance spec- troscopy (EIS) and found that both the oxidation and ad^¬ sorption behaviours of SO₂ depend closely on the potential. More specifically, adsorbed sulfur begins to be oxidized above an applied voltage potential of 0.9 V and can be com^¬ pletely oxidized with a CV maximum applied voltage poten- tial of 1.05 V or higher. US 2006/0249399 Al describes a method for in-situ cleaning of fuel cell electrodes, more specifically PEM fuel cell electrodes, said method being performed by applying a power pulse, using a low-power supply, across the fuel cell elec- trodes. The power pulse removes chemisorbed chemical spe^¬ cies, such as SO₂, from the electrochemical catalyst of the electrodes .

It is an object of the present invention to provide a means for improving the regeneration of the average cell potential of a sulfur-poisoned fuel cell, fuel cell part, fuel cell stack or fuel cell stack system, especially a solid oxide fuel cell (SOFC) stack. It is an additional feature of the present invention to provide an improved process for the regeneration of the cell potential of the sulfur- poisoned fuel cell, fuel cell part, fuel cell stack or fuel cell stack system, especially the solid oxide fuel cell (SOFC) stack, and to improve the performance thereof to above the previous or initial average cell potential (per^¬ formance) value.

Thus, the present invention relates to a method for regen^¬ eration of a sulfur-poisoned fuel cell, fuel cell part, fuel cell stack or fuel cell stack system, especially a solid oxide fuel cell (SOFC) stack, wherein said method comprises applying a voltage potential, or current pulses as described in US 2006/0249399 Al, to the sulfur-poisoned fuel cell, fuel cell part, fuel cell stack or fuel cell stack system for a period of time sufficient to increase the performance, e.g. the average cell potential, of the sulfur-poisoned fuel cell fuel cell part, fuel cell stack or fuel cell stack system. Besides sulfur, other species may also be removed from the sulfur-poisoned fuel cell, fuel cell part, fuel cell stack or fuel cell stack system, said species being one or more elements selected from C, Si, CI, F, Br, P, Cr, Na, K, Al, Sr, Se, As, Sb, Pb, Hg and Cd.

In a preferred embodiment the method for regeneration of a sulfur-poisoned fuel cell stack results in the regeneration of the anode of the fuel cell stack.

In another embodiment the voltage potential applied to the sulfur-poisoned fuel cell stack ranges from 0.1 to 5 V per cell, preferably from 0.7 to 1.5 V per cell. In a further embodiment the period of regeneration time sufficient to increase the cell potential after regenera^¬ tion in SOFC mode of the sulfur-poisoned fuel cell stack is from about 0 to about 20 hours, preferably from about 0 to about 3 hours, more preferably from about 0 to about 1 hour, most preferably about 30 minutes.

In a still further embodiment the regeneration is initiated after a detected performance loss or before the cell stack is subjected to an electrical load. In this connection, de- tected performance loss could mean any decrease in average cell potential after the cell stack has been subjected to an electrical load for any period of time.

The regeneration can be initiated after any percentage de- crease in average cell potential of the fuel cell stack.

The regeneration can be a full regeneration, a partial regeneration or a full regeneration and an improvement. Full regeneration means that the performance is recovered to a value corresponding to the performance of the fuel cell before the previous drop in performance associated with the fuel cell being subjected to an electrical load and a sulfur-containing fuel or an anode feed gas compris^¬ ing sulfur (e.g. sulfur-containing species H₂S, SO₂, COS, and organic sulfur-containing components) . Figure 1 illus^¬ trates a full recovery. The initial performance, here shown as voltage, e.g. average cell voltage, is the voltage ob^¬ tained after the previous regeneration or the original per^¬ formance. As time passes with the fuel cell operating on a sulfur-containing fuel, the performance degrades as illus^¬ trated by a drop in the voltage. Typically the voltage loss levels off to a relatively stable level. At this point the figure illustrates the application of the regeneration, and the performance obtained afterwards is back to the same level as the initial performance. This series of events can be repeated any number of times. The recovery can be un- derstood as having removed enough sulfur from the anode to restore the performance to the level obtained initially or just after the last regeneration.

Full regeneration and an improvement means that the fuel cell performance is higher than the value observed before the previous performance loss. Figure 2 illustrates a full recovery with improvement. The initial performance, here shown as voltage, e.g. average cell voltage, is the voltage obtained after the previous regeneration or the original performance. As time passes with the fuel cell operating on a sulfur-containing fuel, the performance degrades as il^¬ lustrated by a drop in the voltage. Typically the voltage loss levels off to a relatively stable level. At this point the figure illustrates the application of the regeneration, and the performance obtained afterwards is initially at a level which is higher that the initial performance. This series of events can be repeated any number of times. The recovery can be understood as having removed enough sulfur from the anode to restore the performance and to improve it to the level which is higher than obtained initially or just after the last regeneration.

Partial regeneration means that the performance of a fuel cell stack is increased by less than 100% of the loss in performance observed after the fuel cell stack is first used or less than 100% of the performance obtained after the previous regeneration. Figure 3 illustrates a partial recovery. The initial performance, here shown as voltage, e.g. average cell voltage, is the voltage obtained after the previous regeneration or the original performance. As time passes with the fuel cell operating on a sulfur- containing fuel, the performance degrades as illustrated by a drop in the voltage. Typically the voltage loss levels off to a relatively stable level. At this point the figure illustrates the application of the regeneration, and the performance obtained afterwards is only partially recovered to a lower level than the initial performance. This series of events can be repeated any number of times. The recov^¬ ery can be understood as having removed enough sulfur from the anode to recover some performance, but not enough to reach the level obtained initially or just after the last regeneration. This could also in part be due to a general degradation of the fuel cell stack performance, not neces^¬ sarily caused by sulfur. The above three examples of regeneration, i.e. full regen^¬ eration, full regeneration with improvement and partial re^¬ generation, can be combined in any order and any number of times. Figure 4 illustrates a series of partial regenera^¬ tions vs. initial performance.

A further embodiment of the invention consists in subject^¬ ing the fuel cell stack to a flow of gases during regenera- tion. Flowing a gas through the anode side will facilitate a purge of liberated sulfur species from the anode. A suit^¬ able gas to be used as anode feed gas comprises one or more of the following gaseous components: hydrogen, nitrogen, water (steam), oxygen, carbon monoxide and carbon dioxide. The anode feed gas during regeneration may be: an inert gas, preferably nitrogen

hydrogen

steam

- a steam-H₂ mixture, possibly with other gases

such as an inert gas (N₂) or mixtures of the above. The anode gas may also contain oxygen, either continuously or in pulses .

In a further embodiment of the invention the anode feed gas flow is about 30 Nl/min/ (100 cm² anode area) or less. Anode area is the geometric area of the anode, e.g. a 10 cm x 10 cm anode will have an anode area of 100 cm². During regeneration it is preferred that the anode feed gas of N₂ is between about 0 and about 10 Nl/min/ (100 cm anode area), more preferably the anode feed gas of 2 is about 5

Nl/min/ (100 cm² anode area) . Preferably, the anode feed gas of ¾ is between about 0 and about 2 Nl/min/ (100 cm² anode area) , more preferably the anode feed gas of ¾ is about 0.21 Nl/min/ (100 cm² anode area) . Preferably, the anode feed gas of ¾0 is between about 0 and about 5 Nl/min/ (100 cm² anode area) , more preferably the anode feed gas of ¾ is about 2 Nl/min/ (100 cm² anode area) .

Further the cathode may either have no flow during regeneration or have a flow of e.g. air. The flow rate of air may be reduced as compared to what is used in normal SOFC oper^¬ ation, e.g. in order to control the stack temperature.

The following parameters may be monitored during the use of the fuel cell stack, e.g. power generating mode of the fuel cell stack (s), or during stand-by mode:

Voltage loss: Once a certain performance loss (depending on operating conditions such as current density, sulfur concentration, fuel utiliza^¬ tion etc.) since the last regeneration (or initial operation) has been realized, then a new regeneration cycle is initiated. The voltage loss can be defined as a fraction of the full loss po^¬ tential or the full potential at which point a steady state is reached. Figure 1 illustrates re^¬ generation after reaching close to steady state, while figure 5 illustrates regeneration after reaching some performance loss which is less than the expected steady state performance loss. Absolute voltage: Once a certain lower voltage limit is reached, a new regeneration cycle is in^¬ itiated .

Rate of voltage change: Once the rate of change in voltage has dropped below a certain lower limit, then a new regeneration cycle is initiated. The rate of change is measured over a period of time long enough to establish a stable value, the trend of which can be monitored.

Regeneration is initiated after a certain amount of time has passed since the last regeneration. This can be abso^¬ lute time, time at temperature, time operating in fuel cell mode, time weighted by the amount of fuel flow passing the anode (a representation of the amount of sulfur entering the anode) or a suitable combination of these methods.

The above parameters may be used individually or together to indicate a suitable period for regeneration of the fuel cell, a suitable duration and conditions for the regenera^¬ tion or sequence of regeneration steps.

The regeneration of the fuel cell may occur at any point in time. If the fuel cell stack system is used intermittently (e.g. a truck APU system), then the regeneration could conveniently take place: during start-up of the fuel cell stack system, during stand-by of the fuel cell stack system, where the system is kept warm

during shut-down of the fuel cell stack system and during operation as needed, and

when operating, load can be supplied by another source, e.g. a battery, and the stack can be re^¬ generated .

Once the load is removed from the fuel cell stack system and it is no longer operating in power generating mode, the anode regeneration gas can be supplied to the anode and a suitable gas supplied to the cathode (e.g. air), and the regeneration can be initiated.

The power supply/voltage potential for regeneration may come from: - a power supply unit, e.g. EA-PSI 8080-60 2U 0-80V

0-60A,

batteries, e.g. charged in parallel during normal stack operation and used for regeneration (e.g. by connection in series) ,

- subset of batteries, e.g. charged (e.g. in paral^¬ lel) during normal stack operation and used for regeneration, e.g. by connection in series, or capacitors, e.g. charged in parallel during nor^¬ mal stack operation and used for regeneration by connection in series main engine generator (APU system), e.g. through a voltage-increasing device .

Furthermore, the power supply/voltage potential for regen- eration may: be produced by the main engine power generation device and voltage-adjusted to suit regeneration, or

come from another SOFC stack or stacks or system still operating in power generating mode. The anode off-gas from a system operating this way could be used as regeneration gas.

The regeneration may be carried out very frequently, for example every few seconds for a very short period of time. In this case the stack may remain connected to the power load during regeneration. The power connection has a low pass filter, and the regeneration pulse is applied in high frequency. In this way the stack will continue to produce power while being regenerated.

The gas is preferably hydrogen, and hydrogen and water are both supplied externally. Another possibility is that the water is supplied externally, while the hydrogen is gener- ated internally by electrolysis, or that the feed comprises both hydrogen and water, but more hydrogen is generated internally by electrolysis. Some other feed gases, such as AdBlue (which is a 32.5 wt% solution of urea in demineral- ized water normally used to reduce emission of nitrogen ox- ides in exhaust gases from heavy diesel vehicles) , will al^¬ so work. Anode exhaust from one or more other SOFC stacks can also be used as a regeneration gas, especially if a possible sulfur content is reduced or virtually eliminated, e.g. by a sulfur removal device such as a ZnO material or a Ni material. Preferably the regeneration gas has a residence time of be^¬ tween 1 ms and 10 s. The feed gas comprises 2-80 %, prefer^¬ ably 10-60 % water, and 40-100 % hydrogen. Generally the feed gas comprises CO (0-100 %) , C0₂ (0-100 %) , H₂0 (0-100 %), N₂ (0-100 %), CH₄ (0-100 %) and H₂ (0-100 %) . The gas may contain a portion of anode exhaust from another stack system and additional H₂ is generated internally via elec^¬ trolysis .

Furthermore, the regeneration gas may contain a portion of the outlet of a secondary steam reformer for methane, propane or LPG (liquefied petroleum gas) , a liquid fuel re^¬ former, a clean coal pyrolyzer, a syngas generator or a truck exhaust with a cleaning section.

Besides sulfur, other species may also be removed from the sulfur-poisoned fuel cell, fuel cell part, fuel cell stack or fuel cell stack system, said species being one or more elements selected from C, Si, CI, F, Br, P, Cr, Na, K, Al, Sr, Se, As, Sb, Pb, Hg and Cd.

The method according to the invention can be carried out in a once-through system as shown in figure 6. The system comprises an optional power supply.

If the power supply is not used, the feed stream and the product gas stream will have approximately the same gas composition. The product gas will contain sulfur species, and thus sulfur will be removed from the stack as long as the feed gas has a lower sulfur content. Preferably the feed gas contains no sulfur or at least only a negligible amount of sulfur. If the above-mentioned AdBlue is used as feed, it will de^¬ compose to approximately 54.2 wt% water, 9.2 wt% CO₂, 9.2 wt% ₂ and 27.5 wt% ¾ . This is a good regeneration gas, but not the most preferred composition. By applying power using the optional power supply, it is possible to achieve more desirable gas compositions. Table 1 below as well as Fig. 2 illustrate the stack exit gas composition as a func^¬ tion of the conversion of ¾0 in the stack by electrolysis.

Table 1

If the feed was water, the relationship would look as shown on Fig. 3 and indicated in Table 2 below. Table 2

Other feeds can be used as well. In choosing a feed it is advantageous to have a relatively high hydrogen content and a suitable water concentration as discussed for preferred compositions, because then a good regeneration throughout the stack is provided. If water is supplied externally, while the hydrogen is generated internally by electrolysis, the hydrogen concentration will be very low at the inlet and the regeneration will be slower than if the feed had some hydrogen. But as the gas passes through the stack, the concentration of hydrogen will increase, making the regen- eration faster.

To regenerate fast it is advantageous to have a high flow rate through the stack. If the exit gas reaches the equi^¬ librium sulfur concentration, then doubling the flow rate will mean doubling the regeneration speed. This, however, results in a high consumption rate of the feed material, and if electrolysis is used to modify the gas composition, it will result in a high power consumption also. However, by introducing a recycle system, both can be lowered quite significantly while maintaining a high regeneration speed. Figure 7 shows an example of a recycle based system. The gas leaving the SOFC stack exit will contain sulfur and possibly be in equilibrium with the adsorbed sulfur in the stack. A direct recycling of the gas will therefore not re- suit in any further sulfur removal. It is thus necessary to remove all the sulfur or a part thereof from the recycle gas. This can be done by passing the recycle gas through a sulfur removal device such as an adsorption bed capable of adsorbing sulfur. Said adsorption bed must contain a suita- ble adsorption material, such as a Ni based reforming cata^¬ lyst, a ZnO based material or the like.

This adsorption bed can be placed in the system in many ways :

(a) As shown in Fig. 7 as part of a dedicated route used during regeneration, while under normal SOFC operation the bed is bypassed by the anode recycle typically used. (b) In the anode recycle loop, thus always exposed to gas flow, both in regeneration mode and in typical SOFC operation mode with anode recycle.

(c) In the inlet stream to the SOFC stack.

(d) A recycle flow (not shown) typically operates at a low^¬ er temperature than the SOFC exit stream, and thus a cool^¬ ing heat exchanger is applied. The adsorption bed can be placed either before or after said heat exchanger, but preferably in a position where the temperature is optimal for adsorption, in which case several heat exchangers may be used. Furthermore, by using heat exchangers it is possi- ble to place valves controlling the recycle flow paths in a cold position, thereby eliminating the need for hot valves.

(e) As the adsorption bed will eventually be saturated, it is desirable to place it in a position for easy replacement and potentially have it connected with some sort of quick connects for low cost maintenance.

The invention is illustrated further by the following exam- pies.

Example 1

This example illustrates a test in which sulfur is removed from a solid oxide fuel cell stack by applying a current to the stack. The figure shows the results of the test. The following abbreviations are used:

DR = Diesel Reformate (15.4% H₂, 13.7% CO, 9.8% C0₂, 11.3% H₂0, 49.8% N₂)

DR* = 45% H₂, 20.2% N₂, 13.7% CO, 9.8% C0₂, 11.3% H₂0.

ME = Method Embodiment

H₂S = Hydrogen Sulfide A = Amperes (Current) Start up = Start up of test Shutdown = Shut down of test All examples of the present invention are carried out using an 11 cell SOFC stack comprising NiYSZ based anode support^¬ ed cells with an YSZ (yttria-stabilised zirconia) electro^¬ lyte and LSCF (lanthanum, strontium, cobalt, ferrite) based cathode and with a footprint of 12 x 12 cm². The Topsoe

Fuel Cell reference is: TOFC stack K-652 11 Cell. Examples of suitable SOFC stacks are provided in WO 2011/137916. The examples are carried out as a series of sequential steps (a to v) on the same cell stack over time, as illustrated in Figure 8. Steps a to v in Figure 8 correspond to method steps provided in Examples 2-8.

The power supply used throughout the examples was an: EA- PSI 8080-60 2U 0-80V 0-60A.

For all examples the stack is placed in a furnace with a temperature of about 700 °C and the feed gases are heated to about 700°C. The cell stack potential is measured as an an^¬ ode feed gas composition of either DR or DR* is fed to the anode with a flow of 11 Nl/min, and to the cathode with a flow of 28 Nl/min, and a current (electrical load) is drawn from the cell stack. The cell stack is then subjected to sulfur poisoning by addition of H₂S to the anode feed gas composition. After the cell stack is poisoned with sulfur, the cell stack is subjected to the anode feed gas composi^¬ tions and conditions of the present invention, followed by measurement of the cell stack potential.

An anode feed gas composition of DR means an anode feed gas composition of: 15.4% H₂, 13.7% CO, 9.8% C0₂, 11.3% H₂0, 49.8% N₂. An anode feed gas composition of DR* means an anode feed gas composition of: 45.0 % H₂, 13.7% CO, 9.8% C0₂, 11.3% H₂0, 20.2 % N₂. The current drawn from the cell stack may be either 20A or 17A. @20A denotes a current of 20A drawn from the cell stack. @17A denotes a 17A current drawn from the cell stack . The anode feed gas compositions and conditions may be ab^¬ breviated in the following form:

DR@20A conditions means an anode feed gas composition of DR (15.4% H₂, 13.7% CO, 9.8% C0₂, 11.3% H₂0, 49.8% N₂) , an an- ode feed gas flow of 11 Nl/min to the anode, and 28 Nl/min to the cathode, and a current of 20A is drawn from the cell stack .

DR*@20A conditions means an anode feed gas composition of DR* (45.0 % H₂, 13.7% CO, 9.8% C0₂, 11.3% H₂0, 20.2 % N₂) , an anode feed gas flow of 11 Nl/min to the anode, and 28 Nl/min to the cathode, and a current of 20A is drawn from the cell stack. DR@17A conditions means an anode feed gas composition of DR (15.4% H₂, 13.7% CO, 9.8% C0₂, 11.3% H₂0, 49.8% N₂) , an an^¬ ode feed gas flow of 11 Nl/min to the anode, and 28 Nl/min to the cathode, and a current of 17A is drawn from the cell stack . Example 2 a) The average cell potential was measured to be about 720 mV under DR@20A conditions. b) Sulfur was added to the anode feed gas (DR@20A condi^¬ tions) as ¾S, such that the concentration of sulfur reached 2 ppm of the volume of the feed gas composi^¬ tion. The average cell potential dropped to about 700 mV, showing a 20 mV decrease in cell stack potential due to the addition of the sulfur. c) The electrical load was disconnected from the SOFC stack and the anode feed gas was changed to 2 (5 Nl/min) . A power supply was connected to the stack and it was operated for about half an hour with the ap^¬ plied voltage potential varying from 0.77 to 1 V per cell in a cyclic manner. Subsequently the anode feed gas was changed to ¾ (0.21 Nl/min) and ¾0 (2 ml/min) and the applied voltage potential was 0.9 V per cell for about half an hour. d) The cell stack potential was measured under DR@20A

conditions to be about 725 - 730 mV. This shows that the process applied in step c) resulted in a full re^¬ covery of the 20 mV cell potential lost in step b) . The performance of the cell was even improved by 5-10 mV above the original reference measurement in step Example 3 [continued from step d) ] e) The cell stack of step d was used in step e) . Sulfur was added to the anode feed gas (DR@20A conditions) as ¾S, such that the concentration of sulfur reached 2 ppm of the volume of the feed gas composition. The av^¬ erage cell potential dropped to about 710 mV, showing a 20 mV decrease in cell potential due to the addition of the sulfur. f) The electrical load was disconnected from the SOFC

stack and the anode feed gas was changed to ¾ (0.21 Nl/min) and ¾0 (2 ml/min) . A power supply was connected to the cell stack and it was operated for about 1½ hour with the applied voltage potential of 0.98 V per cell. Subsequently the gas was changed to 2 (5 Nl/min) and the stack was operated for about half an hour with the applied voltage potential varying from 0.79 to 1 V per cell in a cyclic manner. g) The cell stack potential was measured under DR@20A

conditions to be about 745 mV. This shows that the process of the invention applied in step f) resulted in a full recovery of the 20 mV cell potential that was lost in step e) , due to sulfur poisoning of the anode. The performance was even improved by about 25 mV above the original reference measurement in step Example 4 [continued from step g) ] h) The cell stack of step g) was used in step h) . Sulfur was added to the anode feed gas (DR@17A conditions) as ¾S, such that the concentration of sulfur reached 20 ppm of the volume of the feed gas composition. After poisoning the cell with sulfur, the average cell po^¬ tential dropped to about 705-710 mV measured at 17A, showing a 40 mV drop compared to step g) . It should be noted that step h) is measured at 17A and step g) is measured at 20A, therefore, the loss of cell potential is larger than 40 mV. [See steps j) and k) for a comparison of cell potentials measured under 17A and 20A current conditions] . i) The electrical load was disconnected from the SOFC

stack and the anode feed gas was changed to ₂ (5 Nl/min) and ¾0 (2 ml/min) . A power supply was connected to the stack and it was operated with an ap- plied voltage potential of 1.2 V per cell. Subsequent^¬ ly the gas was changed to ¾ (0.21 Nl/min) and ¾0 (2 ml/min) and the stack was operated with an applied voltage potential of 1.2 V per cell. The anode feed gas was then changed to pure ₂ and the stack was op- erated with an applied voltage potential of 1.0 V per cell. The total time of this step was about 4 hours. j) The cell stack potential was measured under DR@20A

conditions to be about 738mV. This shows that the pro- cess of the invention applied in step i) resulted in a partial recovery [compared to 745 mV measured in step g) ] of the reduction in cell potential seen in step h) , due to sulfur poisoning of the anode. The performance was even improved by about 25 mV above the orig^¬ inal reference measurement in step a) . Partial recov^¬ ery means that the cell potential has been restored to a value higher than cell potential before applying the process of the invention, but lower than the highest cell potential possible. k) The cell stack potential was measured under DR@17A

conditions; the cell potential was about 778mV. The cell stack potential measured under DR@17A conditions corresponds to the potential measured in step g) under DR@20A conditions. When the cell stack potential meas^¬ ured under DR@17A conditions is compared to the cell potential reduction observed in step h) , it shows that the loss in cell potential due to sulfur poisoning in step h must have been about (778-708) = 70 mV.

1) The SOFC stack was shut down for some days

Example 5

) The cell stack of step 1) was used in step m) . The

SOFC stack was restarted and again the cell stack po tential was measured under DR@20A conditions to be about 725 mV. This value is greater than the cell po tential obtained in step a) , but less than what was measured in step j) prior to shut down.

In order to operate the stack with high concentrations of sulfur the feed gas was changed to DR* under DR*@20A conditions. The performance was measured to be about 780 mV.

Step m) illustrates two cell potential measurements: the first where the gas feed composition is DR and the cell potential measurement is 725 mV, i.e. under

DR@20A conditions; and the second where the gas feed composition is DR* and the cell potential measurement is 780 mV, i.e. under DR*@20A conditions. These two measurements illustrate a difference of 55 mV in cell potential between the two gas feed compositions DR and DR* under the same gas flow and current conditions. n) Sulfur was added to the anode feed gas (DR*@20A condi- tions) as ¾S, such that the concentration of sulfur reached 200 ppm of the volume of the composition gas feed. The average cell potential dropped to about 750 mV from 780 mV, showing a 30 mV reduction in cell potential due to the addition of the sulfur. o) The electrical load was disconnected from the SOFC

stack and the anode feed gas was changed to ₂ (2-5 Nl/min) . A power supply was connected to the stack and it was operated with an applied voltage potential of 1.2 V per cell for about half an hour. Subsequently, the gas was changed to ₂ (5 Nl/min) and ¾0 (1.6 ml/min) and the stack was operated with an applied voltage potential of 1.0 to 1.3 V per cell for about half an hour to one hour. p) The cell potential was measured under DR@20A condi^¬ tions to be about 720 mV. 720 mV under DR@20A condi- tions corresponds to approximately 775 mV under

DR*@20A conditions. This shows that the process in step o) resulted in a partial recovery of the cell po^¬ tential; i.e. from 750 mV in step n) to 775 mV in step o) under DR*@20A conditions. The cell potential is now the same as in step a) and in step m) under

DR@20A, but significantly lower than what was obtained after step g) (745 mV) and step i) (738 mV) , where the potential was improved to about 740-745 mV per cell.

Example 6 [continued from step p) ] q) The cell stack of step p) was used in step q) . Sulfur was added to the anode feed gas (DR*@20A conditions) as ¾S, such that the concentration of sulfur reached

200 ppm of the volume of the composition gas feed. The average cell potential decreased to about 750 mV from the calculated 775 mV cell potential of step p) , due to the addition of the sulfur. This reduction in cell potential corresponds to the reduction in cell poten^¬ tial observed in step n) . The corresponding calculated cell potential value under DR@20A conditions is 695 mV.

) The SOFC stack was operated under DR@20A conditions for about 20 hours. The performance improved very slowly, similar to the performance recovery illustrat ed in figure 1, and stabilized at a new level. s) The cell potential was measured under DR@20A condi^¬ tions to be about 730 mV. This shows that the 20 hours of operation with a sulfur free fuel (DR@20A condi- tions) could partially regenerate the performance of the cell stack compared to the cell stack performance obtained after application of the method in steps f) and i) . However, the recover was very slow, about 20 hours .

Example 7 [continued from step s) ] t) The cell stack of step s) was used in step t) . The

electrical load was disconnected from the SOFC stack and the anode feed gas was changed to ₂ (5 Nl/min) and H₂O (2 ml/min) and the stack was operated with an applied voltage potential of 1.1 V per cell for about 20 hours. u) The cell potential was measured under DR@20A condi^¬ tions to be about 740-745 mV. This is an improvement over the cell potential obtained in step s) and illus^¬ trates that the process of the present invention can improve the performance above what is possible by sim^¬ ple purging the SOFC with sulfur-free fuel. The per^¬ formance has now reached the same level as was ob^¬ tained after steps f) and i) . This illustrates that the process of the present invention can consistently improve the performance of the SOFC above the original performance and that the reduction in cell potential suffered by adding respectively: 2 ppm, 20 ppm and 200 ppm ¾S to the feed gas composition can be completely recovered. In addition, the recovery period under the conditions of the present invention are significantly faster than the recovery period observed under known conditions, i.e. purging with sulfur-free fuel and re^¬ sults in a better performance.

Example 8 v) The cell stack of step u) was used in step v) . The

cell potential of the SOFC was measured under DR*@20A conditions. The performance was 776-780 mV per cell, which corresponds to the performance obtained in step m) , verifying the complete cell potential recovery un^¬ der DR*@20A conditions

Example 9

In a recycle system based on AdBlue the following is an ex^¬ ample of achievable gas compositions as a function of recy^¬ cle split (the fraction of the stack exit that is recycled to the inlet rather than being purged from the loop) .

This example is based on keeping a fixed flow rate to the stack and aiming at obtaining a fixed conversion of 85 % of the steam entering the stack. The example clearly shows the advantage of increasing the recycle. The feed material con- sumption and the power consumption for electrolysis both drop essentially linearly towards zero as the recycle split increases towards 100 %.

In addition this example clearly shows that as the recycle increases, then a more uniform gas composition throughout the stack is achieved, as the inlet gas composition and the exit gas composition approach each other more and more. In table 3 below, SR is the split to recycle (%) , Rff is the relative feed flow (%) and RP is the relative power (%) .

Table 3

Example 9 is shown graphically in Fig. 9. Example 10

This example shows the effect of increasing the conversion of the feed stream at a constant recycle split of 95 % us^¬ ing AdBlue as the feed material. The conversion is shown as two values :

1. conversion vs. feed = water converted / water in feed (Val.l) and

2. conversion vs. stack feed = water converted / water in feed to stack (Val.2); see the following table.

Table 4

Example 10 is illustrated graphically in Fig. 10, which shows the mole fraction in the exit gas (in percent) as a function of the ¾0 conversion by electrolysis vs feed.

A similar relationship can be determined for any given feed (AdBlue, water, mixtures etc.) to determine which operating condition will minimize the power consumption while obtaining gas compositions which will facilitate regeneration.

Claims

Claims :

1. A method for regeneration of a sulfur-poisoned fuel cell, fuel cell part, fuel cell stack or fuel cell stack system, especially a solid oxide fuel cell (SOFC) stack, wherein said method comprises applying a voltage potential to the sulfur-poisoned fuel cell stack for a period of time sufficient to increase performance, e.g. the average cell potential of the sulfur-poisoned fuel cell stack.

2. A method according to claim 1, wherein the voltage potential applied per cell ranges from 0.1 to 5 V, preferably from 0.7 to 1.3 V.

3. A method according to claims 1 and 2, wherein the pe^¬ riod of time required to improve the performance of the fuel cell stack is from about 0 to about 20 hours, prefera^¬ bly from about 0 to about 3 hours, more preferably from about 0 to about 1 hour.

4. A method according to claims 1 to 3, wherein the anode and cathode of the fuel cell are supplied with suitable feed gases.

5. A method according to claims 1 to 4, wherein the anode feed gas flow is about 30 Nl/min/ (100 cm² anode area) or less .

6. A method according to claims 1 to 5, wherein the anode feed gas comprises at least 0 - 100% hydrogen.

7. A method according to claims 1 to 5, wherein the anode feed gas comprises at least 0 - 100% nitrogen.

8. A method according to claims 1 to 5, wherein the anode feed gas comprises at least 0 - 100% water.

9. A method according to claims 1 to 5, wherein the anode feed gas comprises at least hydrogen and water.

10. A method according to claims 1 to 5, wherein the anode feed gas comprises a gas suitable for facilitating regener^¬ ation.

11. A method according to claims 1 to 10, wherein the re- generation is initiated after a detected performance loss or before the cell stack is subjected to an electrical load .

12. A method according to claims 1 to 11, wherein the re- generation is full regeneration, partial regeneration or full regeneration and an improvement.

13. A method according to claims 1 to 12, wherein during the regeneration the fuel cell stack remains connected to the power load.

14. A method according to claims 1 to 13, wherein the an^¬ ode of the fuel cell stack is regenerated.

15. The method according to any of the preceding claims, where other species are also removed from the sulfur- poisoned fuel cell, fuel cell part, fuel cell stack or fuel cell stack system, said species being one or more elements selected from C, Si, CI, F, Br, P, Cr, Na, K, Al, Sr, Se, As, Sb, Pb, Hg and Cd.

16. Use of a potential voltage for removal of sulfur from a fuel cell stack.