Production of hydrogen with low CO x -content for PEM fuel cells by cyclic water gas shift reactor

ARTICLE IN PRESS

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/ijhydene

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 3 5 4 – 1 3 6 0

0360-3199/$ - see frodoi:10.1016/j.ijhyden

�Corresponding auFax: +49 391 6110 35

E-mail address: s

Production of hydrogen with low COx-content for PEM fuelcells by cyclic water gas shift reactor

Vladimir Galvitaa, Torsten Schrodera, Barbara Mundera, Kai Sundmachera,b,�

aMax Planck Institute for Dynamics of Complex Technical Systems, Sandtorstr. 1, 39106 Magdeburg, GermanybOtto von Guericke University, Process Systems Engineering, Universitatsplatz 2, 39106 Magdeburg, Germany

a r t i c l e i n f o

Article history:

Received 1 February 2007

Received in revised form

5 October 2007

Accepted 11 December 2007

Available online 28 January 2008

Keywords:

Water gas shift reaction

CO removal

Iron redox cycle

Carbon deposition

PEM fuel cell

nt matter & 2008 Internae.2007.12.022

thor. Max Planck Institute3.undmacher@mpi-magde

a b s t r a c t

Hydrogen gas with low CO content was produced by cyclic water gas shift (CWGS) reactor

based on the periodic reduction and re-oxidation of Fe2O3–CeO2–ZrO2. The process was

operated with CO/H2 mixtures produced by e.g. auto-thermal reforming of hydrocarbons.

During the reduction phase of the cyclic process, the incoming CO/H2 mixture converted

Fe2O3–CeO2–ZrO2 into a reduced form. Subsequently, steam was fed into the reactor for re-

oxidation of the reduced material. Thereby, H2 was released which can be used for a proton

exchange membrane fuel cell (PEMFC) without any further purification. As side product,

some coke can be formed on the solid surface by Bouduard reaction. This coke is removed

in the re-oxidation step with steam leading to the formation of carbon monoxide. The

extent of coke formation is controllable by keeping the oxygen conversion of the material

below a certain degree. The feasibility of the novel process was demonstrated by combining

the CWGS reactor with a 5-cell PEMFC stack.

& 2008 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights

reserved.

1. Introduction

Fuel cell development has seen remarkable progress in the

past decades because of an increasing need for enhanced

energy conversion efficiency and because of serious concerns

about the environmental consequences of using fossil fuels

for electricity production. A fuel cell directly transforms

chemical energy in the form of hydrogen into electrical

energy without limitations of the Carnot efficiency. Fuel cell

systems operating on pure hydrogen produce only water, thus

eliminating all emissions locally. The demands for the purity

grade of the used hydrogen fuel are dependent on the type of

fuel cell being considered [1,2]. High-temperature fuel cells

tolerate high concentrations of COx (CO, CO2) in the hydrogen

feed, while this ability is weak for low-temperature fuel cells

because CO adsorbs irreversibly on the surface of the

electrode catalysts, such as Pt, and blocks the reaction

sites for hydrogen. Thus, for the proton exchange membrane

tional Association for Hy

for Dynamics of Compl

burg.mpg.de (K. Sundma

fuel cell (PEMFC), which is a candidate for the propulsion

of vehicles and for dispersed power plants, CO is a strong

poison even at low concentrations. The current state of

PEMFC development requires a hydrogen gas quality of about

yCOo20 ppm.

Conventional hydrogen production technologies such as

steam reforming, auto-thermal reforming and partial oxida-

tion of methane yield large amounts of CO as by-product.

Reduction of the CO content down to the ppm-range using

these processes mostly leads to complex multi-step reaction/

purification trains of bulky apparatuses [2–4]. These draw-

backs remain a serious technological obstacle in the practical

utilization of these processes for PEMFC hydrogen supply.

As a novel alternative to the just mentioned conventional

technologies, methane steam reforming based on the iron

redox cycle, which was designed to convert hydrocarbons to

hydrogen with a quality that exceeds the requirements of all

types of fuel cells, has obtained significant attention [5–14].

drogen Energy. Published by Elsevier Ltd. All rights reserved.

ex Technical Systems, Sandtorstr. 1, 39106 Magdeburg, Germany.

cher).

ARTICLE IN PRESS

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 3 5 4 – 1 3 6 0 1355

This two-step process is performed in a single reactor unit

without any additional post-processing of the product gas, i.e.

no additional water gas shift reactors and/or preferential

oxidation reactors are needed. The technology is based on the

periodic reduction/re-oxidation of iron oxides. During the first

step of the cycle (iron oxide reduction), methane (or gasified

biomass) reduces the iron oxide to iron, Eq. (1). During the

second step (iron re-oxidation), steam is used as oxidizing

agent for iron, simultaneously producing hydrogen according

to Eq. (2). The produced gas consists of steam and pure

hydrogen that can be supplied directly to a PEMFC:

CH4 þ Fe3O4 ! CO2 þ 2H2Oþ 3Fe, (1)

4H2Oþ 3Fe! 4H2 þ Fe3O4. (2)

However, the overall hydrogen production depends on the

slower process step, which is the reduction of iron oxide to

iron, Eq. (1). Furthermore, the iron oxides which are carrying

the redox reaction can be deactivated by sintering phenom-

ena. Otsuka et al. [13] and Takenaka et al. [14] have found that

the addition of Cr cations to iron oxides mitigated the

sintering of iron metal and/or iron oxides during the redox

process at temperature above 1023 K. Addition of Ni to iron

oxide enhanced the reduction with methane and the

subsequent oxidation with water vapor. The iron oxide

containing both Ni and Cr species was able to produce pure

hydrogen at a temperature of 923 K [13,14].

The reaction rate of reduction can be enhanced and the

deactivation of iron oxides can be reduced by applying the

concept of a periodically operated two-layer catalytic reactor,

as suggested in a previous work [15]. In that reactor, two fixed

beds of different catalytically active oxide materials are

working at the same temperature. The use of Pt–Ce0.5Zr0.5O2

in the first catalyst layer leads to a significant enhancement of

the methane oxidation rate compared to iron oxides, and thus

allows decreasing of the operating temperature resulting in

an improved overall process performance. Furthermore, the

addition of ceria oxide to iron in the second layer protects

the iron metal and/or iron oxides against sintering during the

Fe3O4

FeOx→Steam:

H2O

Feed Gas:

CO, H2,

CO2, H2O, N2

Switch-over

Valve

Packed be

Fe3O

Phase I:

Phase II:

CH

4

au

toth

erm

al re

form

ing

CH4,

H2O,

air

Fig. 1 – Proposed flow scheme of a fuel processing system for

followed by CWGS reaction.

redox cycle and increases significantly the reduction rate by

H2 and the re-oxidation rate by H2O.

The major disadvantage of the proposed two-layer reactor

is the fact that the first layer is operated periodically for

syngas production although it could be operated under

steady-state conditions with a much higher productivity. This

leads to an improved process concept being based on the

combination of a steady-state reformer followed by a cyclic

water gas shift (CWGS) reactor (see Fig. 1). The advantages of

this modification are a better flexibility and an improved

performance of the overall process. The objective of the

present study is to analyze the behavior of the CWGS reactor

using syngas as feed. In the following, thermodynamic data

as well as kinetic experiments (at Fe2O3–CeO2–ZrO2 as oxygen

storing material) are presented which are an important

prerequisite for the appropriate design of a technical-scale

H2-processor for PEMFC.

2. Experimental

2.1. Preparation of oxide materials

In this study, 30 wt%-Fe2O3–CeO2–ZrO2 was investigated as

catalytically active oxygen storage material for the proposed

cyclic redox process. This material was prepared via urea

hydrolysis [15–17]. Fe–Ce–Zr mixed oxide samples were

synthesized from FeðNO3Þ3 � 9H2O (99.0%, Fluka), CeðNO3Þ3 �

6H2O (99.0%, Fluka) and ZrOðNO3Þ3 � 6H2O (99.0%, Fluka). The

starting metal salts were dissolved in distilled water to the

desired concentration (0.1 M). The ratio between the metal

salts can be altered depending on the desired concentration.

Here we used Fe0.3(Ce0.5Zr0.5O2)0.7. The mixed metal salt

solution was added to a 0.4 M solution of urea (99.0%, Fluka)

with a salt to urea solution ratio of 2:1 (v/v). This mixture was

kept at 100 �C for 24 h. After this step, the sample was allowed

to cool to room temperature prior to being centrifuged in

order to separate a gel product from the solution. The gel was

washed with ethanol, dried overnight in an oven at 110 �C,

→FeOx

Fe3O4 Exhaust Gas:

CO2, H2O, N2

Main Product Gas

for PEMFC:

H2, H2O

Switch-over

Valve

ds filled with

4-CeZrO2

Reduction

Re-Oxidation

PEMFCs based on auto-thermal methane steam reforming

ARTICLE IN PRESS

400 500 600 700 800 900 1000 1100-140

-120

-100

-80

-60

-40

-20

0

20

40

60

Gib

bs f

ree

en

erg

y,

ΔG /

[ k

J/m

ol ]

3Fe2O3 + H2 = 2Fe3O4 + H2O

-80

-60

-40

-20

0

20

40

60

ee

en

erg

y,

ΔG /

[ k

J/m

ol ] FeO + CO = Fe + CO2

Temperature, T / [°C ]

FeO + H2 = Fe + H2O

Fe3O4 + H2 = 3FeO + H2O

2CO = C + CO2

Fe3O4 + CO = 3FeO + CO2

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 3 5 4 – 1 3 6 01356

grinded to a size of about 1mm and then calcinated at 850 �C

for 4 h. Finally, the material was pressed, grinded and sieved.

The resulting mean particle size was 0.25 mm. A more

detailed characterization of the fresh and used oxygen

storage material (30 wt%-Fe2O3–CeO2–ZrO2) can be found in

our previous works [18,19].

2.2. Apparatus and procedure

Redox cycle experiments of 30 wt%-Fe2O3–CeO2–ZrO2 were

carried out in a differential quartz tube reactor (i.d. 10 mm).

Typically, a mass of 10 g of material was packed between

layers of quartz wool. The reactor was placed in an electric

furnace. The temperatures in the center of the packed bed

and at the reactor wall were measured by K-type thermo-

couples. The reduction step (phase I of the process cycle) was

performed by gas mixtures containing carbon monoxide and

hydrogen in helium as carrier gas. The total flow rate of the

feed gas into the reactor was kept constant at 120 ml/min by

Brooks mass flow controller. The reactor was operated in the

temperature range from 700 to 825 �C.

Before re-oxidation, the mixture of rest hydrogen and

carbon monoxide in the lines was removed by helium purging

for 10 min. Subsequently, the re-oxidation of the reduced

material with steam (phase II of the process cycle) was

performed at a flow rate of 120 ml/min immediately after the

reduction of the 30 wt%-Fe2O3–CeZrO2 material. The concen-

tration of the reagents and products was monitored using an

MS Agilent 5973 Network Mass Selective Detector.

The conversion degree of oxygen of the reduced material,

XO2 , was calculated on the basis of the amount of oxygen

which theoretically could be removed from the material and

the amount of oxygen which was released as gaseous

products (H2O, CO2):

XO2¼

NH2O þ 2NCO2

4NFe3O4 þ 0:5NCeO2

� 100%. (3)

The PEMFC stack which was connected to the CWGS reactor

in this study had five cells with a self-breathing cathode and

an cell area of 26 cm2. The anode gas flow rate was 120 ml/min

and the gas inlet temperature was ambient. The membrane

electrode assembly consisted of a NAFIONTM N-105 mem-

brane (DuPont, USA) onto which catalyst layers were applied.

The catalyst used was pure unsupported platinum black of

type HiSPEC1000TM (AlfaAesar-Johnson Matthey GmbH, Ger-

many). The catalyst loading was 1 mg Pt cm�2. In the catalyst

layers also a 10 wt% content (relative to the platinum weight)

of the NAFIONTM polymer was present. The stack was

designed, manufactured and assembled in-house, i.e. by the

workshop of the authors’ institute.

400 500 600 700 800 900 1000 1100

-140

-120

-100

Gib

bs f

r 3Fe2O3 + CO = 2Fe3O4 + CO2

Temperature, T/ [°C ]

Fig. 2 – Gibbs free energy of reduction reactions of several

iron oxides versus temperature. Above: reduction with

hydrogen, below: reduction with CO and Bouduard reaction.

3. Results

3.1. Selection of operating temperature

The optimal reaction temperature for the CWGS process

range can be identified from a detailed analysis of the

thermodynamic data of the reactions being involved. Plots

of the Gibbs free energy change for iron oxides reduction by

CO and by H2 versus temperature are depicted in Fig. 2,

respectively. The data show that the reduction of Fe2O3 to

Fe3O4 and of Fe3O4 to FeO are favored at higher temperatures.

This is valid for H2 as well as for CO. Furthermore, one can see

that—at standard conditions—the formation of carbon by the

Bouduard reaction is thermodynamically not favored at

temperatures above 700 �C.

Coke being produced on the solid material surface during

the reduction cycle phase by the Bouduard reaction, Eq. (4),

leads to a CO production during the re-oxidation step

according to Eq. (5):

2CO2Cþ CO2, (4)

CþH2O2COþH2. (5)

In order to design a stable hydrogen production process, it

is important to identify operation conditions at which carbon

deposition can be avoided. Such conditions can be predicted

by a thermodynamic equilibrium analysis of coupled gas and

solid phases (oxides and carbon) as shown e.g. by Koh et al.

ARTICLE IN PRESS

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 3 5 4 – 1 3 6 0 1357

[20]. Applying the methodology of these authors, in the

present study thermodynamic calculations were done for a

gas composition containing initially 20 vol% CO in helium. As

possible gas phase species CO and CO2 were considered and

as relevant solid species Fe2O3, Fe3O4, FeO, Fe and carbon. The

thermodynamic data were obtained from [21], and for carbon

(here assumed: graphite) from [22]. By minimizing the total

Gibbs energy of the entire system (assuming ideal gas

behavior and treating the solid species as pure condensed

phases), the equilibrium composition of the gas phase and

the solid phase as well as the amount of deposited carbon at

equilibrium was determined at various temperatures and at

different initial oxygen loading of the oxide material.

Fig. 3 shows the calculated degree of oxygen conversion at

which carbon deposition is thermodynamically favored at

given temperature. The analysis reveals that in the tempera-

ture range from 600 to 900 �C at oxygen conversions lower

than XO2¼ 25% elementary carbon is not a stable component

in this system. Coke formation is likely at temperatures below

716 �C if the oxygen conversion degree is higher than

XO2¼ 30%, and over the whole temperature range if the

oxygen conversion degree is close to 100%.

Hence, thermodynamic analysis shows that the tempera-

ture range between 720 and 850 �C is most suitable for

hydrogen production by the proposed redox cycle of the

Fe2O3–CeZrO2 material.

3.2. Hydrogen production

Fig. 4 shows the changes of CO, CO2, H2 and H2O concentra-

tions as a function of time during the reduction phase using

syngas with 40% hydrogen and 15% CO with helium as carrier

gas. The solid material initially was pre-conditioned by

several cycles such that is was fully converted into 30 wt%-

Fe3O4–CeZrO2 [15]. In the beginning, only CO2 and H2O were

0.4600

650

700

750

800

850

900

Te

mp

era

ture

, T

/ [

°C]

Oxygen conversion degree, XO2

Fe2O3 Fe3O4 FeO Fe

1.00.90.80.70.60.50.30.20.10.0

Carbon free area

Carbon deposition area

Fig. 3 – Thermodynamic map of carbon deposition in

dependence of oxygen conversion degree of iron oxides

and temperature.

detected in the product stream. After a reaction time of about

120 s, hydrogen and carbon monoxide were also found in the

outlet gas mixture. Their concentrations increased steadily

and the concentration of CO2 and H2O decreased already

before the oxide material was reduced completely.

Additional reduction experiments with the 30 wt%-Fe3O4–

CeZrO2 material were carried out with a CO/H2 mixture

containing a small amount of methane ðo1 vol% Þ. This is

relevant for the combined operation of a reforming reactor

and the CWGS reactor where the reformer effluent might

contain traces of non-converted methane. With this feed

composition, we observed a methane conversion degree of

20–25% in the initial period of the reaction. The conversion

decreased with the time and achieved 5–7%, but it increased

after the degree of oxygen conversion was higher than 80%.

This observation indicates that methane decomposition

ðCH4 ! 2H2 þ CÞ was catalyzed by reduced iron species. Low

catalytic activity for methane oxidation by iron oxides was

also observed in our previous work [15].

During the addition of steam into the CWGS reactor pure

hydrogen is produced and iron oxides and ceria oxides are re-

oxidized for the next reduction phase of the redox cycle. Fig. 5

shows the observed evolution of the hydrogen formation rate

as a function of time during the re-oxidation of the reduced

material with steam at 800 �C. As can be seen in Fig. 5, the

formation rate of hydrogen during the first 9 min was nearly

constant. Finally, it was possible to repeat the here reported

hydrogen production behavior during 15 redox cycles. During

those cycles the oxygen storage material (30 wt%-Fe3O4–

CeZrO2) lost 17% of its initial activity. The BET surface

area of Fe2O3–CeO2–ZrO2 was found to decrease from

11 m2/g for the fresh sample to 6 m2/g for the used sample.

The mean particle diameters for the fresh and used samples

were 50 and 140 nm, respectively. Thus, deactivation of the

oxygen storage material during redox cycles happened due to

sintering.

0 5 10 15 20 250

10

20

30

40 yH2 = 40 vol. % yCO = 15 vol. %

mcat = 10 g

T = 800 °CF= 120 ml/min

CO2

CO

H2O H2

Co

nce

ntr

atio

n o

f o

utle

t g

as,

yi /

[ v

ol.%

]

Reducing gas

Time, t / [ min ]

F F

Fig. 4 – Evolution of H2, CO, CO2 and H2O concentrations

during reduction phase of Fe3O4–CeZrO2 material with

H2/CO feed mixture using helium as carrier gas.

ARTICLE IN PRESS

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 3 5 4 – 1 3 6 01358

For evaluating the process performance, the following fuel

efficiency Z is defined:

Z ¼NH2 ;out

NH2 ;in þNCO;in¼

NH2O;out þNCO2 ;out

NH2 ;in þNCO;in. (6)

Fig. 6 shows the calculated efficiencies in dependence on the

oxygen conversion degree at an operating temperature of

T ¼ 800 �C. These results clearly indicate that the efficiency

decreases during the reduction of the 30 wt%-Fe3O4–CeZrO2

material and it approaches an approximate value of 30% at

the highest obtained oxygen conversion. At 750 and 825 �C the

efficiency was 25% and 37.5%, respectively. Thus, higher

working temperatures favored higher efficiencies. A tempera-

ture increase results in a better redox ability of the oxygen

0 10 20 30 40 50 60 70 80 90 10020

30

40

50

60

70

80

90

100

30 wt.% Fe3O4-CeZrO2

mcat = 10 g

T = 800 °CyH2

= 40 vol. %F

yCO = 15 vol. %F

F = 120 ml/min

Fu

el e

ffic

ien

cy,

η /

[ %

]

Oxygen conversion degree, XO2 / [ % ]

Fig. 6 – Fuel efficiency calculated from by the ratio of heating

value of generated hydrogen and the heat value of the input

gas as a function of the degree of oxygen conversion of the

applied Fe3O4–CeZrO2 material.

0 5 10 15 20 250

200

400

600

800

1000

1200

1400

1600

yH2O = 75 vol. %

30 wt.% Fe3O4-CeZrO2

mcat = 10 g

T = 800 °CF = 120 ml/min

H2

H2 p

rod

uctio

n r

ate

, r H

2 /

[μm

ol/ m

in]

Oxidizing gas

Time, t / [min]

F

Fig. 5 – Evolution of hydrogen production rate during the re-

oxidation phase of Fe3O4–CeZrO2 material with water vapor

using helium as carrier gas (after preceding reduction with

H2/CO mixture).

storage material. On the other hand, it would be desirable to

operate at as low process temperatures as possible in order to

minimize the rate of iron oxide sintering. Most important, the

process efficiency increases by decreasing the oxygen con-

version level, by increasing the oxygen storage capacity of the

solid phase and by increasing the rate of oxygen release from

the solid Fe3O4–CeZrO2 particles during the reduction phase.

In our previous work [18] it was found that the reduction of

30wt%-Fe3O4–CeZrO2 is a complex transformation process

which includes the simultaneous reduction of the com-

pounds Fe2O3, Fe3O4 and CeZrO2 which are dominated by

different physico-chemical sub-processes. At oxygen conver-

sion below 50% the reduction of Fe2O3 to Fe3O4 and of CeO2 to

Ce2O3 are fast steps being controlled by phase-boundary-

controlled and nucleation-controlled reactions. At higher

oxygen conversion, the reduction of Fe3O4 and CeO2 is

controlled by solid phase diffusion [18].

The obtained results indicate that hydrogen can be

generated with high reproducibility using 30 wt%-Fe3O4–

CeZrO2 as oxygen storage material by reduction with syngas

followed by subsequent oxidation with water vapor. The

purity of the produced hydrogen gas during the iron redox

process is a crucial parameter for its applicability as feed for

polymer electrolyte fuel cells. Fig. 7 shows the evolution of the

CO concentration in the hydrogen product gas during the re-

oxidation step by water vapor at different oxygen conversion

degree. The data indicate that the amount of undesired

carbon monoxide being formed according to the reactions in

Eqs. (4) and (5) depends on the oxygen conversion degree of

the solid material. The conversion should be below 60% in

order to restrict carbon deposition on the surface to an

acceptable level and thereby to guarantee a CO content below

20 ppm in the produced hydrogen gas.

Furthermore, after the reduction investigations, the reactor

was purged for 5 min with He at a flow rate of 240 cm3/min to

clean the gas lines from traces of CO and H2. Subsequently,

He was introduced at a flow rate of 60 cm3/min to burn the

carbon being deposited on the solid surface by using the

oxygen being stored in the Fe3O4–CeZrO2 material according

to the following equation:

CþO� ! CO, (7)

where O� stands for oxygen species from the 30 wt%-

Fe3O4–CeZrO2 material.

Traces of CO and CO2 were quantified by MS. At complete

conversion of oxygen no CO and CO2 traces were detected,

because no free oxygen was available for carbon oxidation.

But as long as the oxygen conversion was restricted to values

lower than 60%, the experiments proved that carbon being

formed on the solid surface can be removed with the help of

oxygen being stored in the Fe3O4–CeZrO2 material. Thus,

there are two ways to suppress carbon formation: (1) by

changing the process conditions such that a lower degree of

oxygen conversion is attained, and/or (2) by developing

carbon-resistant materials.

The overall efficiency of the CWGS process can be defined in

terms of reaction rate (activity) and lifetime (stability) of the

oxygen storage material. For the possible commercialization

of this process, the material deactivation will be the most

important aspect. A systematic investigation of the factors

ARTICLE IN PRESS

0 10 20 30 40 50 603.0

3.5

4.0

4.5

Ce

ll vo

lta

ge

of

the

fu

el ce

ll sta

ck,

U/

[V]

Current density, i / [mA/cm2]

Fig. 8 – Steady-state current–voltage characteristics of a

5-cell PEMFC stack fed with hydrogen gas being produced by

re-oxidation of Fe3O4–CeZrO2 with water vapor at 800 �C.

Fuel cell operating conditions: T ¼ 25 �C, p ¼ 1 bar, anode:

F ¼ 120 ml=min; cathode: ambient air, self-breathing.

0 2 4 6 8 10 12 140

10

20

30

40

50

60

yH2O = 75 vol. %F

30wt.% Fe3O4-CeZrO2

mcat = 10 g

T = 800 °CF = 120 ml/min

XO2 = 30%

XO2 = 60%

XO2 = 80%

CO

co

nce

ntr

atio

n,

yC

O /

[p

pm

]

0 2 4 6 8 100

100

200

300

400

500

XO2 −−> 100%

30wt.% Fe3O4-CeZrO2

mcat = 10 g

T = 800 °CF = 120 ml/min

CO

co

nce

ntr

atio

n,

yC

O /

[p

pm

]

yH2O = 75 vol. %F

Time, t / [min]

Time, t / [min]

Fig. 7 – Evolution of CO concentration during the re-

oxidation with water vapor at different oxygen conversion

degrees resulting from the preceding reduction phase.

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 3 5 4 – 1 3 6 0 1359

influencing the material lifetime can be found in a compa-

nion publication [23]. In general, iron oxides deactivate fast

due to the material sintering during repeated reduction and

re-oxidation cycles. The addition of Ce0.5Zr0.5O2 as promoter

leads to a significantly higher initial activity as well as to a

higher stability [15,19,23]. However, further improvements of

the stability of the oxygen storage material have to be

achieved in the future in order to operate the CWGS process

on a technical scale.

Finally, in the present study the CWGS reactor was

combined with a fuel cell stack. Fig. 8 shows the recorded

steady-state current–voltage curve for a 5-cell PEMFC stack

operated at ambient temperature and pressure. The anode

compartment was fed with hydrogen being produced during

the re-oxidation step of the 30 wt%-Fe3O4–CeZrO2 material at

800 �C. The cathode compartment was fed with air. The

degree of oxygen conversion of 30 wt%-Fe3O4–CeZrO2 by

syngas (40 vol% H2, 15 vol% CO) was kept below XO2o60%.

Concentrations of H2O and He in the produced H2 anodic feed

stream were 2.2 and 10 vol%, respectively. The open circuit

cell voltage of the investigated systems was around

U0 ¼ 4:5 V. A maximal current density of 54 mA/cm2 in the

fuel cell stack was achieved at U ¼ 3:2 V. During the operation

an electric DC power of 4.5 W was generated. The fuel cell

stack was operated over 10 redox cycles (overall 60 min for re-

oxidation steps). After the experiments with the CWGS

reactor, the fuel cell stack was tested at reference conditions

with pure humidified hydrogen feed. No degradation of the

power generation ability of the stack were observed.

4. Conclusions

Fuel processing represents a very important aspect of fuel cell

technology. The widespread utilization of proton exchange

membrane fuel cell (PEMFC) will be possible if CO-free

hydrogen producing technologies are available. The experi-

ments presented in this work confirm the feasibility of the

here proposed cyclic water gas shift (CWGS) reactor being

based on iron redox cycling. This reactor offers a high

flexibility in terms of syngas mixtures which can be converted

into highly pure hydrogen. The CWGS reactor can be

combined with any reformer which yields syngas of a certain

H2/CO-ratio which depends on the type of feedstock being

used in reforming (hydrocarbons, alcohols and biomass).

The CWGS was successfully coupled with a lab-scale 5-cell

PEMFC stack. For the operation of a 1 kW PEMFC stack, we will

now work toward the realization of a pilot-scale CWGS

process. There, further technical tasks will be an efficient

heat management for balancing the endothermic reduction

step against the exothermic re-oxidation of 30 wt%-Fe3O4–

CeZrO2, a further improvement of the stability of the oxygen

storage material by doping of certain components, and the

ARTICLE IN PRESS

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 3 5 4 – 1 3 6 01360

further increase of the volume-specific oxygen storage

capacity in order to keep the reactor size as small as possible.

Acknowledgments

Funding of this research work by the German federal state of

Saxony-Anhalt within the joint project ‘‘Dezentrales brenn-

stoffzellenbasiertes Energieerzeugungssystem fur den statio-

naren Betrieb in der Leistungsklasse 20 kW’’ and by the Max

Planck Society within the joint research project ‘‘ProBio’’ is

gratefully acknowledged.

R E F E R E N C E S

[1] Song C. Fuel processing for low-temperature and high-temperature fuel cells: challenges, and opportunities forsustainable development in the 21st century. Catal Today2002;77(1–2):17–49.

[2] Steele BCH, Heinzel A. Materials for fuel-cell technologies.Nature 2001;414(15):345–52.

[3] Choudhary TV, Goodman DW. CO-free fuel processing for fuelcell applications. Catal Today 2002;77(1–2):65–78.

[4] Armor JN. The multiple roles for catalysis in the productionof H2. Appl Catal A Gen 1999;176(2):159–76.

[5] Fukase S, Suzuka T. Residual oil cracking with generation ofhydrogen: deactivation of iron oxide catalyst in the steam-iron reaction. Appl Catal A Gen 1993;100(1):1–17.

[6] Hacker V, Faleschini G, Fuchs H, Fankhauser R, Simader G,Ghaemi M, et al. Usage of biomass gas for fuel cells by the SIRprocess. J Power Sources 1998;71(1–2):226–30.

[7] Hacker V, Fankhauser R, Faleschini G, Fuchs H, Friedrich K,Muhr M, et al. Hydrogen production by steam–iron process.J Power Sources 2000;86(1–2):531–5.

[8] Hacker V. A novel process for stationary hydrogen produc-tion: the reformer sponge iron cycle (RESC). J Power Sources2003;118(1–2):311–4.

[9] Takenaka S, Son VTD, Hanaizumi N, Nomura K, Otsuka K.Production of pure hydrogen from methane mediated by theredox of Ni- and Cr-added iron oxides. In: Proceedings of the13th International Conference on Catalysis. Paris, France,11–16 July 2004.

[10] Rossini S, Cornaro U, Mizia F, Malandrino A, Piccoli V,Sanfilippo D, et al. Hydrogen with intrinsic CO2 sequestration:

the ENI ‘‘one step hydrogen’’ process DGMK-conferenceinnovation in the manufacture and use of hydrogen.Dresden, Germany; 15–17 October 2003, p. 41–7.

[11] Otsuka K, Yamada C, Kaburagi T, Takenaka S. Hydrogenstorage and production by redox of iron oxide for polymerelectrolyte fuel cell vehicles. Int J Hydrogen Energy2003;28(3):335–42.

[12] Takenaka S, Nomura K, Hanaizumi N, Otsuka K. Storage andformation of pure hydrogen mediated by the redox ofmodified iron oxides. Appl Catal A Gen 2005;282(1–2):333–41.

[13] Otsuka K, Kaburagi T, Yamada C, Takenaka S. Powergeneration using a mesoscale fuel cell integrated with amicroscale fuel processor. J Power Sources2003;122(1–2):111–8.

[14] Takenaka S, Hanaizumi N, Son V, Otsuka K. Production ofpure hydrogen from methane mediated by the redox of Ni-and Cr-added iron oxides. J Catal 2004;228(2):405–16.

[15] Galvita V, Sundmacher K. Hydrogen production frommethane by steam reforming in a periodically operatedtwo-layer catalytic reactor. Appl Catal A Gen2005;289(1–2):121–7.

[16] Pengpanich S, Meeyo V, Rirksomboon T, Bunyakiat K.Catalytic oxidation of methane over CeO2–ZrO2 mixed oxidesolid solution catalysts prepared via urea hydrolysis. ApplCatal A Gen 2002;234(1–2):221–33.

[17] Subrt J, Bohacek J, Stengl V, Grygar T, Bezdicka P.Uniform particles with a large surface area formed byhydrolysis of Fe2(SO4)3 with urea. Mater Res Bull1999;34(6):905–14.

[18] Galvita V, Sundmacher K. Redox behavior and reductionmechanism of Fe2O3–CeZrO2 as oxygen storage material.J Mater Sci 2007;42:9300–7.

[19] Galvita V, Sundmacher K. Cyclic water gas shift reactor(CWGS) for carbon monoxide removal from hydrogen feedgas for PEM fuel cells. Chem Eng J 2007;134(1–2):168–74.

[20] Koh JH, Kang BS, Lim HC, Yoo YS. Thermodynamic analysis ofcarbon deposition and electrochemical oxidation of methanefor SOFC anodes. Electrochem Solid-State Lett2001;4(2):A12–5.

[21] NIST Chemistry WebBook, hhttp://webbook.nist.gov/chemistry/i.[22] Barin I. Thermochemical data of pure substances. Weinheim:

VCH; 1995.[23] Galvita V, Hempel T, Lorenz H, Rihko-Struckmann LK,

Sundmacher K. Deactivation of modified iron oxide materi-als in the cyclic water gas shift process for CO-free hydrogenproduction. Ind Eng Chem Res 2007, in press, doi: 10.1021/ie0708879.