Solar Energy 79 (2005) 409421

www.elsevier.com/locate/solener

Solar water splitting for hydrogen production

with monolithic reactors
C. Agraotis a, M. Roeb b, A.G. Konstandopoulos a,*, L. Nalbandian a,
V.T. Zaspalis a, C. Sattler b, P. Stobbe c, A.M. Steele d
a

Chemical Process Engineering Research Institute, Center for Research and TechnologyHellas (CERTH/CPERI),
P.O. Box 361, 57001 Thermi-Thessaloniki, Greece
b
Deutsches Zentrum fur Luft- und Raumfahrt e.V. (DLR), Institut fur Technische Thermodynamik, Solarforschung,
D-51170 Koln, Germany
c
Stobbe Tech Ceramics (STC), Vejlemosevej 60, DK-2840, Holte, Denmark
d
Johnson Matthey Fuel Cells Centre, Sonning Common, RG4 9NH Reading, UK
Received 21 July 2004; received in revised form 12 November 2004; accepted 23 February 2005
Available online 4 June 2005
Communicated by: Associate Editor Sixto Malato-Rodrguez

Abstract
The present work proposes the exploitation of solar energy for the dissociation of water and production of hydrogen
via an integrated thermo-chemical reactor/receiver system. The basic idea is the use of multi-channelled honeycomb
ceramic supports coated with active redox reagent powders, in a conguration similar to that encountered in automobile exhaust catalytic aftertreatment.
Iron-oxide-based redox materials were synthesized, capable to operate under a complete redox cycle: they could take
oxygen from water producing pure hydrogen at reasonably low temperatures (800
C) and could be regenerated at temperatures below 1300
C. Ceramic honeycombs capable of achieving temperatures in that range when heated by concentrated solar radiation were manufactured and incorporated in a dedicated solar receiver/reactor. The operating
conditions of the solar reactor were optimised to achieve adjustable, uniform temperatures up to 1300
C throughout
the honeycomb, making thus feasible the operation of the complete cycle by a single solar energy converter.
 2005 Elsevier Ltd. All rights reserved.
Keywords: Water-splitting; Solar; Redox materials; Iron oxide; Honeycomb reactors; Hydrogen

1. Introduction

Corresponding author. Tel.: +30 2310 498192; fax: +30 2310

498190.
E-mail address: agk@cperi.certh.gr (A.G. Konstandopoulos).

The harnessing of the huge energy potential of solar

radiation and its eective conversion to chemical energy
carriers such as hydrogen is a subject of primary technological interest. One of the reactions with tremendous
economical impact because of the low value of its
reactants is the dissociation of water (water splitting) to

0038-092X/$ - see front matter  2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2005.02.026

410

C. Agraotis et al. / Solar Energy 79 (2005) 409421

oxygen and hydrogen. The integration of solar energy

concentration systems with systems capable to split water
is of immense value and impact on the energetics and
economics worldwide; by some is considered as the most
important long-term goal in solar fuels production to cut
hydrogen costs and ensure virtually zero CO2 emissions
(Rubbia, 2003; Kodama, 2003; Steinfeld, 2005).
The dissociation of water is a reaction not favoured
thermodynamically; one has to go up to extremely high
temperatures (>2200
C) for obtaining some signicant
dissociation degree (Kodama, 2003). Moreover, direct
one-step water splitting (thermolysis) requires the energy
intensive process of high temperature oxygenhydrogen
separation coupled with expensive membrane technology (Kogan, 2000) and therefore is considered of little
chance for technical and economical viability in the near
future (Perkins and Weimer, 2004). The current state of
the art of solar chemistry for water splitting and hydrogen production is focused on the so-called redox pair cycles. These are two-step processes, based on redox
materials that can act as eective water splitters at lower
temperatures. According to this idea, in the rst step
(water-splitting) the activated redox reagent (usually
the reduced state of a metal oxide) is oxidized by taking
oxygen from water and producing hydrogen, according
to the reaction:
MOred H2 O g ! MOox H2 g

During the second step the oxidized state of the reagent is reduced, to be used again (re-generation), delivering some of the oxygen of its lattice according to the
reaction:
MOox ! MOred 12O2

The disadvantage is that a two-step process is required, consisting of a water splitting and a regeneration
(oxygen release) step. The advantage is the production
of pure hydrogen and the removal of oxygen in separate
steps, avoiding the need for high-temperature separation
and the chance of explosive mixtures formation.
Current research eorts have focused on thermochemical cycles based on various types of ferrites (Tamaura et al., 1995, 1998a,b, 1999, 2001; Kaneko et al.,
2001, 2002a,b; Kojima et al., 1996; Kodama et al.,
2003), iron oxide (Weidenka et al., 1997; Ehrensberger
et al., 1995, 1996), manganese oxide (Sturzenegger and
Nuesch, 1999), zinc oxide (Steinfeld, 2002) and combinations of the above (Steinfeld et al., 1998, 1999). It
has been suggested that the only feasible two-cycle redox
pairs are ZnO/Zn and Fe3O4/FeO (Steinfeld et al., 1999).
The concept has been proven experimentally: with the
most active of these materials, water splitting has reportedly taken place at temperatures as low as 627
C (Tamaura et al., 1998b; Steinfeld et al., 1999), however the
regeneration temperatures are still high (>1300
C), a
most important problem since it can cause signicant

sintering of the metal oxide. Attempts to tackle this

problem have been recently realized by supporting the
redox reagent on high-temperaturestable ZrO2 ne particles (Kodama, 2003; Kodama et al., 2003). In addition,
in such systems it is necessary to quench the reduction
products (Zn/FeO respectively) in order to avoid reoxidation, a fact introducing irreversibilities and complexities in large-scale utilization (such as separate packed
bed reactors for the water splitting and for the regeneration step).
It has been shown though, that complete reduction of
an oxide to the respective metal is not necessary: cation
excess (oxygen decient) NiMn spinel ferrites were proven capable for both water splitting and regeneration
(Tamaura et al., 1995, 1998b; Kojima et al., 1996). In
such systems in the rst endothermic step the ferrite
(Ni0.5Mn0.5Fe2O4 for instance in Tamaura et al., 1995)
was thermally activated above 800
C to form an oxygen-decient ferrite Ni0.5Mn0.5Fe2O4
d (where d indicates the oxygen deciency of the spinel); in the second
step the activated ferrite reacted with water below
800
C to form hydrogen and the initial ferrite. It has
been reported that partial substitution of iron in the spinel phase by Mn, Ni or Zn, forms mixed metal oxides of
the type (MxFe1
x)3O4 that are more reducible and require moderate, more workable upper operating temperature (Steinfeld et al., 1998). These ferrites can contain
not only the dopant metals in the divalent state but a signicant portion of iron as well; thus all possible oxidations to the trivalent state can be exploited for oxygen
uptake from water. Therefore, such spinel materials
are considered as very attractive candidates since one
can make use of the percentage of the divalent metal(s)
present in the spinel phase and oscillate during the redox cycle between the activated and non-activated structure without undergoing phase transformations (such as
to the lower-valence metal oxide or to the respective
metal). In addition, further enhancement of their water
splitting capability is expected if they can be synthesized
with a high concentration of oxygen vacancies in their
crystal lattices (lattice defects). Stoichiometric defects
such as oxygen vacancies and their mobility on the oxide
surface are of great importance for redox reactions
involving gases and oxides of metals with multiple oxidation states and are widely exploited in oxygen-storage
materials, the most common example being CeO2 used
in catalytic gas exhaust aftertreatment (Trovarelli,
1996). The enhanced water splitting activity of non-stoichiometric wustite (Fe1
yO) materials has been attributed to the large amount of defect clusters serving as
nuclei for magnetite formation (Weidenka et al.,
1997) and that of ZrO2-supported Mn0.36Fe2.64O4 in
the contribution of oxygen-decient ZrO2 phases (Kodama et al., 2003).
Despite basic research with respect to active redox
pairs, solar reactor concepts have only recently been re-

C. Agraotis et al. / Solar Energy 79 (2005) 409421

ported in the literature. Some (Steinfeld et al., 1999;

Steinfeld, 2005; Kaneko et al., 2004; Palumbo et al.,
2004) are based on particles fed into rotating cavity reactors, concepts that are complicated and costly to operate. In the present work, an innovative solar water
splitting reactor is proposed based on two concepts:
the synthesis of active iron-oxide-based redox pairs
and their incorporation upon multi-channelled monolithic honeycomb structures capable for achieving and
sustaining high temperatures.
Ceramic multi-channelled monolithic honeycomb
reactors oer an attractive alternative to packed beds
when dealing with gaseous reactants at high temperatures; their advantages include thin walls, high geometric
surface area and therefore good gassolid contact, low
pressure drop, good mass transfer performance and ease
of product separation (Heck and Farrauto, 1995). Further benets can be gained from special material properties such as thermal shock resistance and mechanical
strength. With respect to exploitation of solar energy
in particular, the absorbance of silicon carbide (SiC)
coupled with its high thermal conductivity enables the
collection of solar heat and eective heating of the reactant gases inside the honeycomb channels. It has already
been demonstrated that SiC monoliths can act as collectors of solar heat and achieve temperatures in excess of
1100
C (Hoschmidt et al., 2001; Fend et al., 2004).
Provided that the redox pairs to be developed are capable for both water splitting and regeneration at these
temperature levels, the full cycle can be performed with
the redox material immobilized upon the honeycomb
walls (Konstandopoulos et al., 2000, 2005) avoiding
the needs of continuous powder feeding and collection
from the reactor. The proposed concept not only produces hydrogen but employs the use of renewable solar
energy without any CO2 emissions, in an entirely
clean, natural and environmentally friendly way.

411

droplets and subsequently passed through a high-temperature tube furnace where it transforms within a very
short time to the desired powder (Lorentzou et al.,
2004). The two latter methods provide an alternative
way for synthesis of materials with high defects concentration due to their common characteristic of very short
synthesis times and extremely high cooling rates; in addition they can also take place within reactors that can
provide for controlled oxygen partial pressure, therefore
tuning further the products oxygen vacancies
concentration.
Redox material powders synthesized via the methods
above were subsequently laboratory-tested with respect
to water-splitting activity and regeneration capability
in the form of powders as described below. These series
of experiments were focussed on determining the most
promising material compositions, to be thereafter synthesized in larger quantities and used for the coating
of SiC honeycombs for the eventual solar testing. In parallel, the dedicated solar reactor was designed, built and
tested in a solar furnace facility, in order to minimise
temperature gradients within the volume of the honeycomb absorber and develop the proper solar test programme. A detailed description of the experimental
procedure follows.
2.1. Powder synthesis
2.1.1. Solid-state synthesis (SSS)
For the reasons argued above the targeted redox
materials were MnZn- or NiZn-doped ferrites crystallized in the spinel structure. They were synthesized
by the typical mixed oxide route, which includes
powder mixing of component oxides, prering, milling,
spray drying and calcination at high temperatures
(1250
C) according to the scheme (MnZn ferrites):
Fe2 O3 x=3Mn3 O4 1
xZnO
! Mnx Zn1
x Fe2 O4

2. Experimental
In this work, three routes were employed for the synthesis of iron-oxide-based redox materials: in the rst,
the mixed oxides were prepared with conventional
solid-state synthesis via sintering of the respective oxides
in a controlled-atmosphere sintering furnace; after the
calcination step the materials were cooled under controllable oxygen partial pressure so that materials with controllable defect chemistry were obtained (Tsakaloudi et
al., 2004). The second route involved the so-called selfpropagating high-temperature synthesis (SHS or combustion synthesis) reactions between iron metal powder
and oxygen in the presence of the dopant metal oxides
(Agraotis and Zaspalis, 2004). The third route selected
was Aerosol Spray Pyrolysis where a liquid solution of
the metal precursor salts is atomized in a spray of ne

For the synthesis of the respective NiZn ferrites,

Mn3O4 was replaced by NiO. As already mentioned,
during cool down, the MnZn materials were maintained under controlled O2 partial pressure (following
the equilibrium curve of Fe+2/Fe+3) in order to prevent
complete oxidation of Fe+2 to Fe+3, whereas NiZn ferrites were cooled under air atmosphere.
2.1.2. Self-propagating high-temperature synthesis
(SHS)
The same targeted materials were synthesized via
combustion reactions between iron metal powder and
oxygen in the presence of the dopant metal oxides and
of Fe2O3 powder as a thermal ballast/moderator to
control the synthesis temperature, according to the
scheme (Agraotis and Zaspalis, 2004):

412

C. Agraotis et al. / Solar Energy 79 (2005) 409421

2kFe 1
kFe2 O3 x=3Mn3 O4 1
xZnO

1.5kO2 ! Mnx Zn1
x Fe2 O4

Similarly, for the synthesis of the NiZn ferrites,

Mn3O4 was replaced by NiO. The quantities of the
two reactant dopant oxides (ZnO and either Mn3O4 or
NiO) were calculated so that x = 1
x = 0.5. Temperatures in excess of 1000
C were achieved, whereas the
synthesis times were of the order of few seconds. However, the product was not always of the spinel ferrite
phase. The eect of the ratio of reactant Fe powder to
Fe2O3 moderator (2k/1
k) and of the oxygen content determines the phase composition of the products.
Low values of the (fuel/moderator) ratio (k = 0.3) are
not sucient to complete the reaction, intermediate
values (k = 0.5) favour the formation of single-phase spinels (Mn0.5Zn0.5)Fe2O4 whereas a high value (k = 0.9)
favours the formation of the least-oxygen-containing
doped wustite (MnxZnyFe1
x
yO) as the main phase.
Both series of productsspinels and wustiteswere
checked with respect to water splitting activity.
2.1.3. Aerosol spray pyrolysis (ASP)
Aqueous solutions of nitrates of Fe, Zn, Mn and Ni
with the addition of citric acid and ammonia, were
atomized in sprays of ne droplets and subsequently
passed through a high-temperature tube furnace to be
transformed within a short time to the desired powder
according to the scheme:
2FeNO3 3 xMnNO3 2 1
xZnNO3 2
3C6 O7 H8 8NH3 9.5O2
! Mnx Zn1
x Fe2 O4 18CO2 8N2 24H2 O

Similarly to the previous methods, either Mn(NO3)2

or Ni(NO3)2 was used. The ASP experiments took place
in a stainless steel reactor tube, heated between 350 and
1070
C. Air was used as a carrier gas and the residence
time varied between 0.98 and 0.7 s. A lter holder at the
end of the reactor was used to collect the synthesized
particles. Under these synthesis conditions, the synthesized products were always of the spinel crystal
structure.

is decomposed by pulling away some of the redox reagents lattice oxygen. This treatment (activation) reduces the reagent and results in solids very active
towards water splitting. Three dierent types of carbon
containing species were employed: tylose, diesel fuel
and wood chips. Compared to hydrogen or CO reductive means, these are low- or zero-cost waste streams
of typically low organics concentration. The reduction
process was not found sensitive to the activator type.
The rst set of experiments indicated that by adding less
than 10 wt.% of reductant, reduction was not ecient
and only very small quantities of hydrogen were produced; thus subsequent experiments reported below,
were performed with 10 wt.% of reductant. Both activated and non-activated samples were comparatively
evaluated.
2.3. Reagents characterization with respect to
water-splitting
The materials tested were Mn/(Ni)ZnFeO systems
of spinel (Mnx/(Ni)xZn1
xFe2O4) and of wustite (Mnx/
(Ni)xZnyFe1
x
yO) structure as well as the component
oxides Mn3O4, NiO, ZnO and Fe2O3. A special laboratory unit has been set-up for the evaluation of the redox
reagents, described in detail in Nalbandian et al. (2004).
The basic component of the unit is a U-shaped reactor
placed in a temperature programmable furnace. All the
materials were tested for the water splitting reaction by
successive injections of water in a stream of Ar owing
continuously over the heated sample at temperatures
ranging from 750 to 1100
C. At preselected temperatures, known quantities of high purity water were injected to the reactor by using a microliter syringe. The
euent from the reactor was monitored through a mass
spectrometer (MS) detector (Balzers, Omnistar). The
sensitivity of the MS detector to hydrogen was calibrated by injecting a series of at least 10 constant volume
loops of a 2 mol% H2/Ar mixture. The quantities of
unconverted water and of produced hydrogen were calculated, based on the areas of the corresponding MS
injection peaks.
2.4. Regeneration under inert atmosphere

2.2. Reduction step

In order to improve the water-splitting performance
and perhaps achieve reduction at lower temperatures,
a small quantity of a carbon-containing species (activator) was mechanically mixed with the synthesized powders and the mixture was heated at about 800
C in a
laboratory reactor under Ar ow (Nalbandian et al.,
2004). During heating in air the carbon-containing species is oxidized by O2 from air. However, when the activator-containing solid is heated in an inert atmosphere
without traces of oxygen, the carbon-containing species

The most promising water splitting materials were

studied by thermogravimetric analysis (TGA) in order
to predict the eect of high temperature treatment to
the reductionoxidation cycles expected to occur under
actual operating conditions. The temperature was
ramped at 20
C/min up to 800
C, under Ar ow. After
a 10 min period, the ow was switched to an O2 mixture
(5% O2 in He) and an oxidation process took place for
60 min saturating the reagent with oxygen. The ow
was switched again to Ar and the oxidized reagent was
further heated (20
C/min) up to 1300
C, where it re-

C. Agraotis et al. / Solar Energy 79 (2005) 409421

mained for 30 min. During these heating and isothermal

steps a weight loss was observed, attributed to reagents
reduction (O loss). The rst cycle was completed by a
cooling step at 20
C/min back to 800
C followed by a
new introduction of the oxidizing mixture. During the
second O2 introduction, a weight gain was observed,
attributed to re-oxidation of the material. The same
steps were repeated twice in order to investigate the
successive reductionoxidation behaviour of the
materials.
2.5. Set-up and testing of the thermochemical reactor/
receiver
The main task of the intended test facility is the performance and monitoring of solar chemical hydrogen
production by a two-step water splitting process with a
redox system. The testing facility consists of a receiverreactor as the central element capable of absorbing solar
radiation provided by a solar furnace and at the same
time of carrying out the necessary process steps at the re-

413

quired temperatures. The rst step requires the supply of

pure steam, for the oxygen releasing second step the supply of inert gas, i.e. nitrogen, is needed. The requirements with respect to high temperature operation of
two alternating process steps under dierent conditions
as well as the experience of reactor and receiver designing in former projects were taken into account during
drafting, designing, and constructing the test facility nally leading to the reactor design depicted in Fig. 1.
Single-piece extruded multi-channel honeycombs to
serve as catalyst carriers and cast parts to form the reactors housing vessel (Fig. 1b) were manufactured from
re-crystallized silicon carbide (reSiC) by STC Aps, Denmark. The honeycombs have dimensions B144
200 mm, high density of channels (90 cells per square
inch, channel wall thickness 0.8 mm, cell size 2 2 mm,
or 150 cells per square inch, channel wall thickness
0.6 mm, cell size 1.6 1.6 mm), porosity 36% and mean
pore diameter 6 lm (measured by Hg porosimetry). The
honeycombs are xed in the centre of the reactor, which
has a cylindrical symmetry (Fig. 1a). The focus of the

Fig. 1. (a) Front view of receiver-reactor for solar water splitting, (b) extruded reSiC honeycombs (carriers) and cast parts (housing
vessel), (c) position of thermocouples in the absorber.

414

C. Agraotis et al. / Solar Energy 79 (2005) 409421

solar radiation concentrated by the solar furnace is

located on the circular front area of the honeycombs,
having a diameter of about 140 mm. The receiver is surrounded by isolating material and xed in a cylindrical
housing. The housing is slightly pressurised by a weak
stream of nitrogen preventing condensation of water in
the isolation material and emission of product gases.
The front opening is sealed by a quartz window allowing
the entrance of solar radiation and preventing the exit of
process gas. The sealing can optionally be cooled by air
or water. The process gases steam and nitrogen are fed
into the reaction chamber using radial aligned stainless
steel tubes and at the same time sucked by fans in the
o-gas line to keep the pressure as close as possible to
ambient conditions. A steam generator provides the necessary mass ow of 05 kg/h of steam at a temperature
of about 150
C. The steam can optionally be diluted
by nitrogen at the same temperature. The reaction takes
place in a ceramic honeycomb xed in a ceramic cylinder
whereas a ceramic funnel ensures the transfer of the
product gases into an o-gas tube (Fig. 1b). The ogas is cooled by mixing with nitrogen at ambient temperature and blown o after condensation of remaining
steam. The reactor is integrated in a mini-plant including
all necessary peripherals like inlet of feed gas, steam generator, cooling and treatment of the product gas, piping,
devices for measurement of temperature, pressure, solar
ux distribution, as well as a mass spectrometer for the
detection of the produced hydrogen. The whole miniplant is positioned and operated in the operational room
of the solar furnace at DLR facilities, in Cologne, Germany. The temperature measurement was done by thermocouples within the reactor and on the surface of the
housing. Two thermocouples were pushed into the channels of the honeycomb from the rear side (Tcentre and
Tedge) as indicated by Fig. 1(c) in order to monitor the
absorber temperature. Potential falsication of measurement by direct radiation was avoided by clogging the
two channels aected. Additionally for means of safety
the front surface temperature of the quartz window
was monitored by an infrared camera.

3. Results and discussion

3.1. Evaluation of redox material powders with respect
to water-splitting
The results summarized below are presented in two
types, either as total micromoles of hydrogen produced/g solid or as % mole conversion of injected
water versus total micromoles of water injected/g solid. Deactivation of all tested solids (decrease of conversion) was observed upon successive water injections.
In order to keep a common basis, the conversions compared are the initial ones at 800
C.

3.1.1. Initial activityMaterials tested without reduction

Appreciable H2 yields on as-synthesized reagents (i.e
without reduction) were obtained only from three material families, sintered MnZn-ferrites (SSS), some SHS
Mn(Ni)ZnFeO products and ASP-synthesized ferrites. All other material families became active water
splitters only after reduction (mixing with carbon containing species and heated to 800
C under inert
atmosphere).
MnZn-ferrites (SSS): Samples having the chemical
3
formula (Mn0.75 Zn0.16 Fe2
sintered at
0.09 Fe2 O4
1250
C under controlled O2 pressure to ensure oxygen deciency were active water splitters, producing
pure hydrogen, however with very low conversions
(60 lmoles of water injected over 0.19 g of solid
produced 0.6 lmoles of hydrogen i.e. conversion 1%). The activity of those samples is probably
connected to their divalent iron content and since this
was not high, the obtained H2 yields are consequently
small.
SHS products: Two families of as-synthesized materials were distinguished: one of high and one of low
conversion. Materials having the crystal structure of
wustite (Ax 0 By 0 Fe1
x 0
y 0 )O exhibited high water splitting activitya fact expected since they contain a signicant amount of divalent iron. The water
conversion of the best wustite-type materials
reached 30% (60 lmoles of water injected over
0.23 g of solid produced 18 lmoles of hydrogen).
On the other hand, as-synthesized SHS materials
having a spinel structure exhibited low conversions
attributed to a very low divalent iron content
achieved under the particular synthesis conditions.
Further experiments are under way with dierent
kinds and stoichiometries of dopant metals and
less-oxygen-containing atmospheres in order to synthesize spinel phases with higher divalent iron content expected to perform better as water-splitters.
ASP products: Spinel materials synthesized via ASP
demonstrated signicant water splitting activity and
extremely high H2 conversion ranging between 30%
and 81% depending on the dopants stoichiometry
(for the latter case, 60 lmoles of water injected over
0.20 g of solid produced 48.6 lmoles of hydrogen).
In addition, no CO2 was observed in the product
gases, indicating that these materials are prime candidates for water-splitting.

3.1.2. Materials tested after reduction

Iron oxides: Two types of Fe2O3 were tested: commercial Fe2O3 (Merck Cat. No. 103924) (denoted as
Fe2O3(B)), and an Fe2O3 spray-roasted from iron
chloride solutions (denoted as Fe2O3(A)). Maximum

C. Agraotis et al. / Solar Energy 79 (2005) 409421

water conversion obtained was 35% (60 lmoles of

water injected over 0.19 g of solid produced
21 lmoles of hydrogen) with Fe2O3(A). Fe2O3(B)
also had some considerable initial activity, however, much lower than Fe2O3(A) and deactivated
very fast.
Dopant oxides: NiO produced H2, but at relatively
low yields while Mn3O4 gave signicant yields but
was not considered as a strong candidate, because
the produced H2 was mixed with CO and CO2 (coming from the organic-containing species not totally
consumed during the reduction step).
NiZn ferrites (SSS): As-synthesized materials showed
negligible water conversion and H2 production.
However, after reduction, they gave appreciable
activity to produce H2 with conversions reaching
11% (60 lmoles of water injected over 0.2 g of solid
produced 6.6 lmoles of hydrogen).
MnZn-ferrites (SSS): This family was capable for
water splitting even without reduction (but with
low conversion). After reduction, conversion reached
12% (60 lmoles of water injected over 0.2 g of
solid produced 7.2 lmoles of hydrogen).
SHS ferrites: As already mentioned, as-synthesized
MnZn-ferrites were inactive. After reduction the
materials became water splitters, with initial conversion 7% (60 lmoles of water injected over 0.19 g
of solid produced 4.2 lmoles of hydrogen).

3.1.3. Final comparison of candidate water-splitting

materials
A comparison among the materials tested with respect to the total hydrogen yield per gram of solid, is
presented in Fig. 2 as a function of the total micromoles
of H2O injected (but in three separate plots for reasons
of clarity). Table 1 presents the nal comparison of the
materials tested so far. In the rst column the initial
water conversion at 800
C is presented, while the second column shows the total H2 yield per gram of solid,
for the rst 1000 lmoles of injected water. The reduced
Fe2O3(A) presents high initial conversion combined with
relatively high H2 yield. Similar results were obtained
over the reduced SHS MnZn ferrites with somewhat
lower initial conversion. The SHS-produced, wustitetype materials (Mnx/NixZnyFe1
x
yO) gave also high
initial conversion and H2 yields, without reduction.
The reduced SSS MnZn spinel ferrites have also been
proven active. The interesting fact is that spinels synthesized by ASP exhibited the highest conversions so far
(3081% depending on the dopants stoichiometry). An
interesting comparison is between spinels of the stoichiometry (Mn0.5Zn0.5)Fe2O4: these synthesized via SHS
exhibited negligible conversion whereas those by ASP
exhibited a conversion of 29.75%. The reasons for this
discrepancy are currently under investigation; the supe-

415

riority of ASP materials might be due to their much ner

particle size compared to that of the SHS ones (250 nm
versus 25 lm respectively). On the other hand the performance of the ASP-spinel materials further supports
the arguments in the introduction that indeed, spinels
can make use of the divalent metal ions for the splitting
of water and the synthesis of hydrogen. All of the above
materials are active water splitters; for identifying the
best ones, regeneration capability should be also taken into account.
3.2. Evaluation of redox reagents with respect to
regeneration capability
An eective redox pair water splitter has to be able
for regeneration and H2 production after repetitive
reductionoxidation cycles. Fig. 3 shows the weight
and the temperature curves as a function of time for a
sintered MnZn ferrite (synthesized via the SSS method),
according to the procedure presented in Section 2. After
the initial oxidation process at 800
C where the sample
gains weight, a weight loss of 0.68% is observed during
heating under inert gas ow, in the temperature range
8001300
C, with maximum at 1075
C. The same
weight was gained back during the isothermal oxidation
step at 800
C. A second reduction cycle indicated very
similar weight loss during heating, with maximum at
1090
C.
Representative samples of all the important material
families have been tested in the same way and the results
are presented in Table 2. The temperature range within
which each sample exhibited weight loss is reported. It
is obvious that many tested materials can be reduced under inert atmospheres at temperatures well below
1300
C and that the reductionoxidation cycle can be
repeated at least two times. Fe2O3 undergoes a highly
repeatable weight loss-gain cycle, with a relatively high
value of oxygen loss (3.5%); however exhibits the
highestregeneration temperature (1285
C). On the other
hand, SHS-produced materials and sintered ferrites
exhibit much lower regeneration temperatures, even
though with less repeatability than Fe2O3. ASP
(Mn0.5Zn0.5)Fe2O4 exhibited an extremely high oxygen
loss (11%) but at a relatively high regeneration temperature (1260
C) whereas the ASP ZnFe2O4 showed the
opposite trend: very low oxygen loss (12%) but at
much lower temperature (1005
C). All these families
of materials seem most promising for further exploitation. It seems that a trade-o between oxygen regeneration percentage and regeneration temperature is
necessary.
The results corroborate ndings from previous literature (Tamaura et al., 1998b) where ferrites were claimed
to exhibit relatively small amount of oxygen liberated
during activation, but on the other hand lower, more
workable operating temperatures. Taking into account

416

C. Agraotis et al. / Solar Energy 79 (2005) 409421

600

SHS - (Mn0.5Zn0.5)FeO

Total moles H2 produced/ g solid

Total moles H2 produced/ g solid

600

SHS - (Ni0.5Zn0.5)FeO
SHS - (Mn0.5Zn0.5)Fe2O4+Activator

500

SHS - (Mn0.5Zn0.5)Fe2O4
Mn3O4 + Activator
NiO+Activator

400

300

200

100

Fe2O3 (A) +Activator

Fe2O3 (B) +Activator

500

SSS - (Ni0.66Zn0.33)Fe2O4+Activator
SSS - (Mn0.75Zn0.16)Fe2O4+Activator
SSS - (Mn0.75Zn0.16)Fe2O4

400

SSS - (Ni0.66Zn0.33)Fe2O4

300

200

100

0
0

1000

2000

3000

4000

5000

6000

1000

Total moles H2 produced/ g solid

2000

3000

4000

5000

6000

Total moles H2O injected/ g solid

Total moles H2O injected/ g solid

3000

2000

1000

ASP - Mn0.5Zn0.5Fe2O4
ASP - Ni0.2Zn0.8Fe2O4
ASP - ZnFe2O4
0
0

10000

20000

30000

40000

50000

60000

Total moles H2O injected/ g solid

Fig. 2. Total micromoles of hydrogen produced/g solid versus total micromoles of water injected/g solid for representative samples.

the water splitting results mentioned before, the four

families of materials presented in Table 2, seem the most
promising for further exploitation.
3.3. Comparison with respect to maximum possible
hydrogen yield
For a particular reagent composition the maximum
possible H2 yield can be calculated from the water splitting reaction stoichiometry and the maximum possible
oxygen uptake from the water molecules. However,
there was no way to know a priori, the maximum possible oxygen uptake capability in the reagents lattice;
therefore this quantity had to be determined implicitly.
TGA was employed for this determination: by rst
reducing the reagent and then by keeping the reduced reagent under an oxidizing atmosphere for a prolonged
period of time and measuring its weight gain until stabilization, one can measure the maximum amount of oxygen that this particular reagent composition is capable
of uptaking. This quantity of oxygen can then be com-

pared to the amount of oxygen actually uptaken from

water (at a particular water splitting reaction temperature for example) in order to determine the reagents
water splitting eciency. The respective procedure
was as follows: reagent samples of 4080 mg were heated
with 20
C/min up to 800
C under Ar ow. After
10 min the ow was switched to an O2 mixture (5% O2
in He) and the oxidation process took place for
60 min, until no more weight gain of the sample was observed. In Fig. 4a such a TGA run for an SHS
(Mn0.5Zn0.5)FeO material is shown. From the total
weight gain (oxygen uptake in micromoles O2/g solid)
and the water splitting reaction (H2O ! H2 + (1/2)O2)
the maximum expected H2 to be produced (in micromoles H2/g solid) at the specic temperature (800
C)
can be calculated. At the laboratory water splitting unit
0.2 mg of the same material was heated at 800
C,
under Ar atmosphere. Water was injected at steady conditions, until the solid was almost completely deactivated. Fig. 4b shows the hydrogen yield expressed as
total micromoles H2 produced/solid as a function of

C. Agraotis et al. / Solar Energy 79 (2005) 409421

417

Table 1
Final comparison of the most representative materials tested with respect to water splitting
Material

Initial % mole conversion

of injected H2O (800
C)

SHS (Mn0.5Zn0.5)FeO
SHS (Ni0.5Zn0.5)FeO
SHS (Mn0.5Zn0.5)Fe2O4
ASP (Mn0.5Zn0.5)Fe2O4
ASP (Ni0.2Zn0.8)Fe2O4
ASP ZnFe2O4
SHS (Mn0.5Zn0.5)Fe2O4 + activator
3
SSS Mn0.75 Zn0.16 Fe2
0.09 Fe2 O4
2
SSS Mn0.75 Zn0.16 Fe0.09 Fe3
2 O4 activator
SSS (Ni0.66Zn0.33)Fe2O4
SSS (Ni0.66Zn0.33)Fe2O4 + activator
Fe2O3(A) + activator
Fe2O3(B) + activator
NiO + activator
Mn3O4 + activatora

28 3
22 3

Total micromoles H2 produced/g

solid for 1000 lmoles H2O
injected/g solid (typically 0.2 g solid)
200 20
170 20
0
297 30
812 50
592 40
65 10
4 0.5
80 5
0
80 10
270 20
135 10
81
310 20

0
30 3
80 10
65 8
7.5 2
0.75 0.2
11 2
0
10.5 2
34 3
26 3
1.1 0.3
66 8

Error ranges estimated from repetitive experiments.

a
This material gave a small quantity of CO during the experiment.

100.8

total micromoles H2O injected/g solid. The total

hydrogen yield obtained during the experiment was
1111 lmoles H2/g solid, which is 90.8% of the maximum
possible.

1400
Argon flush
Argon flush

100.6

100.4
1000
100.2
O2 mixture
Introduction

Temperature (C)

] Weight (%)

1200

800
100.0
O2 mixture
Introduction

99.8

600
40

80

120

160

200

240

Time (min)

Fig. 3. Weight loss (d) and temperature () curves

obtained during TGA study of a sintered SSS MnZn ferrite.

3.4. Set-up and testing of the thermochemical reactor/

receiver
An experimental test campaign was carried out in the
solar furnace to obtain experimental experience and to
qualify the receiverreactor with regard to the tasks of
operating at temperatures up to 1200
C and to carry
out a thermochemical cycle for solar water splitting.
The central aims of the test campaign were:
test of feasibility of all components, in particular with
regard to temperature resistance against thermal
stress and temperature gradients,

Table 2
Regeneration comparison of the most representative materials tested
Material

Weight loss during

rst reduction (%)

Temperature
range (
C)

Peak
(
C)

Weight loss during

second reduction (%)

Temperature
range (
C)

Peak
(
C)

SHS (Mn0.5Zn0.5)FeO
SHS (Mn0.5Zn0.5)Fe2O4
SSS MnZn ferrite
Fe2O3(A)
Fe2O3(B)
Fe2O3(A) + 10% activator

1.59 0.1
1.19 0.1
0.68 0.1
3.32 0.2
3.31 0.2
3.09 0.2 (3.33 0.2
based on Fe2O3(A))
10.90 0.1
1.20 0.1

8001250
8001210
8001300
12001300
12001300
12001300

1065
1045
1075
1286
1290
1286

8001210
8001210
8001210
11201295
10801300
11101295

1060
1055
1090
1285
1280
1285

8001270
8001285

1260
1005

1.12 0.1
0.79 0.1
0.79 0.1
3.36 0.2
3.35 0.2
3.02 0.2 (3.37 0.2
based on Fe2O3(A))
9.1 0.1
2.07 0.1

8001275
8001285

1280
1095

ASP (Mn0.5Zn0.5)Fe2O4
ASP ZnFe2O4

Error ranges estimated from repetitive experiments.

418

C. Agraotis et al. / Solar Energy 79 (2005) 409421

exhibited a strong gradient from 800
C at the edge to

1250
C in the centre. The gradient decreases within
the absorber due to radial heat transfer. To reduce such
a high gradient, the position of the aperture was shifted
in axial direction causing a distance of 150 mm between
aperture and focus position. This was done by shifting
the reactor along the optical axis towards the concentrator of the solar furnace. By that the ux density in the
aperture plane was equalised and the temperature dierence was diminished to about 100 K near the front surface of the absorber and less within the honeycomb.
Fig. 5 shows the axial temperature prole in the honeycomb (length: 150 mm) after the modication described. The axial prole was determined by stepwise
pulling the thermocouples to the rear part of the honeycomb. An almost constant temperature levelwithout
considering the very rst millimetrescould be achieved
in the complete volume of the absorber after convergence to a thermal equilibrium. A further increase of
that level is conceivable and intended for future. Therefore a maximum of volume and surface is potentially
applicable for the thermochemical cycle, in particular
for the regeneration step during the testing of coated
honeycombs.
3.5. Transfer to a continuous operation

monitoring and enhancement of temperature prole

within the absorber,
inuence of the solar ux distribution in the focus on
the behaviour the absorber structure,
study of inuence of process parameters.
In a rst series of tests the temperature distribution
was monitored using dierent heating ramps and dierent levels of solar input power. The temperature of all
components remained within the acceptable limits, even
after many hours of operation. The only impact was observed at the surface of the radiation duct. Temperatures
of up to 800
C at this position caused discolorations
but without derogation of operation.
All components operated reliably even after several
instantaneous jumps of input power, which can be a sudden start-up or shut down. A crucial point for the performance of the thermochemical cycle is the achievable
temperature level and the temperature distribution within the absorber. In the rst experiments the positions of
the receiver aperture, which is the front face of the honeycomb absorber (see Fig. 1c), and the focal plane of the
solar furnace almost coincided. As a result of that the
temperature of the absorber near the irradiated face

1200

Temperature (C)

Fig. 4. SHS (Mn0.5Zn0.5)FeO material (a) TGA run, full

oxidation, (b) hydrogen yield from water splitting at 800
C.

The operation of a commercial plant will require a

continuous operation enabling a continuous supply of
the main product hydrogen. At the same time each
coated honeycomb will be available only about half of
the time for hydrogen production because during the
second half it is being regenerated. Nevertheless a continuous provision can be ensured by using a modular
set-up of spatially xed honeycomb absorbers in separate reaction chambers on a solar tower.

1000

T(centre)

800

T(edge)
600

Tmax. (quartz window) = 400 C

T exhaust gas = 870 C

400
0

10

12

14

Position (cm)

Fig. 5. Axial temperature distribution in the irradiated honeycomb (sealed reactor, near to thermal equilibrium, errormax(T) =
15
C: resolution limit of data acquisition hardware).

C. Agraotis et al. / Solar Energy 79 (2005) 409421

One part of modules splits water while the rest is

being regenerated. After completion of the reactions
the regenerated modules are switched to the splitting
process and vice versa by switching the feed gas. Simultaneous to that the solar ux will be optically varied in
an alternating way depending on the status of the individual modules. This is necessary for the compensation
of the dierent heat demands and dierent reaction temperatures in both reaction steps. When using ferrites the
regeneration step is endothermic, the splitting step is
slightly exothermic (Tamaura et al., 1998b) a fact that
should decrease the hydrogen yield at higher temperatures, whereas the reaction rate is accelerated with
increasing temperature level. The detailed eect of temperature on hydrogen yield and the optimisation of the
water splitting temperature reaction is the subject of current on-going work. It is certain however, that this
would be below the regeneration temperature of most
of the materials tested (12001300
C). Therefore an
operation at alternating temperature levels will presumably be necessary. Based on the powder experiments and
on the available reaction volume the duration of a single
cycle is estimated as considerably less than 1 h. Therefore several cycles will be operated at one day.
Experimental values of the energy conversion eciency, i.e. the ratio of the energy stored in chemical
form (enthalpy change chemical conversion), to the
solar energy input to the reactor are not yet available
due to the fact that within the scope of this work, only
non-solar experiments of water splitting by the redox
materials as powders have been performed and reported.
The potential eciency of the process in the solar test
reactor depends in one hand on the conversion of water
to hydrogen achieved and in the other hand on the thermal losses of the solar reactor. It should be borne in
mind that currently both these factors are not optimised.
The thermal qualication of the solar test reactor provided mean thermal losses of about 2.3 kW during water
splitting and of about 5.1 kW during regeneration under
steady state conditions. About 1/3 of that amount may
be recoverable because it is excess heat of the product
gases. With the conversion achieved so far (81%) and
a projected mass ow rate of steam in the solar reactor
about 1 kg/h, preliminary calculations based on Tamauras thermodynamic data (Tamaura et al., 1998b) have
shown that the eciency of the solar test reactor (lower
heating value of hydrogen produced/solar energy input)
can reach up to 55%. The eciency will be optimised by
the development of enhanced water-splitting compositions with higher water-to-hydrogen conversion, by
thermally optimising the reactor and in particular by
the scale-up of the process, which reduces the ratio of
surface of the reactor to the reactive zone and thereby
the heat losses. A decrease of the overall eciency will
occur due to the energy required to recycle/separate
the inert gas (such as nitrogen) needed to provide inert

419

atmosphere for the reduction step. Depending on the

size of the plant, pressure swing absorption or membrane methods will be applied to separate the released
oxygen and to purify the nitrogen, methods needing
about 0.25 kW h/Nm3 nitrogen. The exact value depends on the required purity of the nitrogen, which is
still an object of investigation. Preliminary calculations
indicate that, in general, a reduction of eciency of
about one fourth is expected by the need to use pure
nitrogen as ushing gas. This reduction can be counterbalanced as mentioned above, by improvements in the
overall reactor design and operation that can maximize
hydrogen production during the water-splitting step
and minimize heat losses during entire continuous operation. However, a recent investigation (Takahashi et al.,
2004) shows a release of oxygen in similar redox systems
that can reach nearly 100% under air-ow conditions.
This opens the perspective to avoid inert gas completely
or at least minimise the required amount.

4. Conclusions
A new promising method for performing solar heated
thermo-chemical process applying two-step water splitting operating at a moderate temperature level of about
8001200
C has been presented. It includes the combination of a support structure capable of achieving high
temperatures when heated by concentrated solar radiation, with an iron-oxide-based redox system suitable
for the performance of water dissociation and subsequent regeneration at these temperatures. To the
authors knowledge this is the rst time that such a solar
reactor concept based on monolithic honeycombs is
being proposed and tested.
A wide variety of iron-oxide-based redox pair systems was synthesized, characterized and comparatively
evaluated with respect to water splitting and regeneration capability. The proposed concept has been proved:
the synthesized catalyst systems are able to split water
and generate hydrogen as the only product, at temperatures as low as 800
C with conversions reaching 80% in
terms of water (amount of water converted/total amount
of water). From the regeneration experiments, it became
obvious that several of these materials could be reduced
under inert atmospheres at temperatures well below
1300
C and that the reductionoxidation cycle could
be repeated at least two times, indicating high potential
for their integration in a two-step water splitting process
within a solar system. Hydrogen yields more than 90%
of the maximum possible were achieved from at least
one catalyst system synthesized.
A dedicated solar chemical receiver/reactor suitable
to carry out the described two-step thermo-chemical
cycle has been designed, constructed, and qualied. A
test programme has been developed. The temperature

420

C. Agraotis et al. / Solar Energy 79 (2005) 409421

prole within the (B144 200 mm) honeycombs has

been homogenised to minimise temperature gradients
and provide almost equal reaction conditions within
the whole volume of the honeycomb absorber, being
now capable of achieving uniform temperatures of the
order of 1300
C.
The next steps of the work include on one hand further optimisation of the redox materials with respect to
dopants stoichiometry in order to further improve their
hydrogen yield and their capability for regeneration and
on the other hand cyclic testing of both process steps
using redox reagents-coated honeycombs incorporated
in the solar reactor.

Acknowledgments
The work has been funded by the European Commission within the Project Catalytic monolith reactor for
Hydrogen generation from solar water splittingHYDROSOL (ENK6-CT-2002-00629), under the Energy, environment and sustainable development (part b:
Energy) Programme (19982002).

References
Agraotis, C., Zaspalis, V.T., 2004. Self-propagating hightemperature synthesis of MnZn-ferrites for inductor applications. Journal of Magnetism and Magnetic Materials 283
(23), 364374.
Ehrensberger, K., Frei, A., Kuhn, P., Oswald, H.R., Hug, P.,
1995. Comparative experimental investigations of the
water-splitting reaction with iron oxide Fe1
yO and iron
manganese oxides (Fe1
xMnx)1
yO. Solid State Ionics 78,
151160.
Ehrensberger, K., Kuhn, P., Shklover, V., Oswald, H.R., 1996.
Temporary phase segregation processes during the oxidation of (Fe0.7Mn0.3)0.99O in N2H2O atmosphere. Solid
State Ionics 90, 7581.
Fend, T., Hoschmidt, B., Pitz-Paal, R., Reutter, O., Rietbrock, P., 2004. Porous materials as open volumetric solar
receivers: experimental determination of thermophysical
and heat transfer properties. Energy 29 (56), 823833.
Heck, R.M., Farrauto, R.J., 1995. Catalytic Air Pollution
Controlcommercial Technology. Van Nostrand Reinhold, New York, USA.
Hoschmidt, B., Fernandez, V., Konstandopoulos, A.G.,
Mavroidis, I., Romero, M., Stobbe, P., Tellez, F., 2001.
Development of ceramic volumetric receiver technology, In:
Funken, K.-H., Bucher, W. (Eds.), Proceedings of 5th
Cologne Solar Symposium, June 21, Forschungsbericht
2001-10, DLR, Germany, pp. 5161.
Kaneko, H., Hosokawa, Y., Gokon, N., Kojima, N., Hasegawa, N., Kitamura, M., Tamaura, Y., 2001. Enhancement
of O2-releasing step with Fe2O3 in the water splitting by
MnFe2O4Na2CO3 system. Journal of Physics and Chemistry of Solids 62, 13411347.

Kaneko, H., Kojima, N., Hasegawa, N., Inoue, M., Uehara,

R., Gokon, N., Tamaura, Y., Sano, T., 2002a. Reaction mechanism of H2 generation for H2O/Zn/Fe3O4.
International Journal of Hydrogen Energy 27, 1023
1028.
Kaneko, H., Ochiai, Y., Shimizu, K., Hokokawa, Y., Gokon,
N., Tamaura, Y., 2002b. Thermodynamic study based on
the phase diagram of the Na2OMnOFe2O3 system for H2
production in three-step water splitting with Na2CO3/
MnFe2O4/Fe2O3. Solar Energy 72, 377383.
Kaneko, H., Kodama, T., Gokon, N., Tamaura, Y., Lovegrove, K., Luzzi, A., 2004. Decomposition of Zn-ferrite for
O2 generation by concentrated solar radiation. Solar Energy
76, 317322.
Kodama, T., 2003. High-temperature solar chemistry for
converting solar heat to chemical fuels. Progress in Energy
and Combustion Science 29 (6), 567597.
Kodama, T., Kondoh, Y., Kiyama, A., Shimizu, K., 2003.
Hydrogen production by solar thermochemical water-splitting/methane-reforming process. In: Proceedings of International Solar Energy Conference ISEC Hawaii, USA,
1618 March 2003, pp. 121128.
Kogan, A., 2000. Direct solar thermal splitting of water and onsite separation of the products-IV. Development of porous
ceramic membranes for a solar thermal water-splitting
reactor. International Journal of Hydrogen Energy 25,
10431050.
Kojima, M., Sano, T., Wada, Y., Yamamoto, T., Tsuji, M.,
Tamaura, Y., 1996. Thermochemical decomposition of H2O
to H2 on cation-excess ferrite. Journal of Physics and
Chemistry of Solids 57 (11), 17571763.
Konstandopoulos, A.G., Kostoglou M., Skaperdas, E., Papaioannou, E., Zarvalis, D., Kladopoulou, E., 2000. Fundamental studies of diesel particulate lters: transient loading,
regeneration and aging, SAE Trans. 109 (J. Fuels and
Lubricants) Paper SAE Tech. Paper 2000-01-1016. Also in
Diesel Exhaust Aftertreatment 2000, SAE SP-1497 pp. 189
211.
Konstandopoulos, A.G., Papaioannou, E., Zarvalis, D., Skopa,
S., Baltzopoulou, P., Kladopoulou, E., Kostoglou, M.,
Lorentzou, S., 2005. Catalytic lter systems with direct and
indirect soot oxidation activity, SAE Tech. Paper 2005-010670 (SP-1942).
Lorentzou, S., Karadimitra K., Agraotis C., Konstandopoulos A.G., 2004. New routes for ferrite powders synthesis. In:
PARTEC 2004, International Conference for Particle
Technology, Nuremberg, Germany, 1618 March.
Nalbandian, L., Zaspalis, V.T., Evdou, A., Agraotis, C.,
Konstandopoulos, A.G., 2004. Redox materials for hydrogen production from the water decomposition reaction.
Chemical Engineering Transactions 4, 4348.
Palumbo, R., Keunecke, M., Moller, S., Steinfeld, A., 2004.
Reections on the design of solar thermal chemical reactors:
thoughts in transformation. Energy 29 (56), 727
744.
Perkins, C., Weimer, A.W., 2004. Likely near-term solarthermal water splitting technologies. International Journal
of Hydrogen Energy 29 (15), 15871599.
Rubbia, C., 2003. Hydrogen at crossroads between science and
politics. Conference on the Hydrogen EconomyA Bridge
to Sustainable Energy, Brussels, 1617 June.

C. Agraotis et al. / Solar Energy 79 (2005) 409421

Steinfeld, A., 2002. Solar hydrogen production via a two-step
water-splitting thermochemical cycle based on Zn/ZnO
redox reactions. International Journal of Hydrogen Energy
27, 611619.
Steinfeld, A., 2005. Solar thermochemical production of
hydrogena review. Solar Energy 78, 603615.
Steinfeld, A., Sanders, S., Palumbo, R., 1999. Design aspects of
solar thermochemical engineeringA case study: Two-step
water-splitting cycle using the Fe3O4/FeO redox system.
Solar Energy 65 (1), 4353.
Steinfeld, A., Kuhn, P., Reller, A., Palumbo, R., Murray, J.,
Tamaura, Y., 1998. Solar-processed metals as clean energy
carriers and water-splitters. International Journal of Hydrogen Energy 23 (9), 767774.
Sturzenegger, M., Nuesch, P., 1999. Eciency analysis for a
manganese-oxide-based thermochemical cycle. Energy 24,
959970.
Takahashi, Y., Aoki, H., Kaneko, H., Hasegawa, N., Suzuki,
A., Tamaura, Y., 2004. Oxygen-gas-releasing reaction of Zn
ferrite by Xe lamp beam irradiation in air at 1800 K. Solid
State Ionics 172, 8991.
Tamaura, Y., Steinfeld, A., Kuhn, P., Ehrensberger, K., 1995.
Production of solar hydrogen by a novel, 2-step, watersplitting thermochemical cycle. Energy 20 (4), 325330.

421

Tamaura, Y., Hasegawa, N., Kojima, M., Ueda, Y., Amano,

H., Tsuji, M., 1998a. Water splitting with the Mn(II)-Ferrite
CaOH2O system at 1273 K. Energy 23 (10), 879886.
Tamaura, Y., Kojima, M., Sano, T., Ueda, Y., Hasegawa, N.,
Tsuji, M., 1998b. Thermodynamic evaluation of water
splitting by a cation excessive (Ni, Mn) Ferrite. International Journal of Hydrogen Energy 23, 11851191.
Tamaura, Y., Ueda, Y., Matsunami, J., Hasegawa, N., Nezuka,
M., Sano, T., Tsuji, M., 1999. Solar hydrogen production
by using ferrites. Solar Energy 65 (1), 5557.
Tamaura, Y., Kojima, N., Hasegawa, N., Inoue, M., Uehara,
R., Gokon, N., Kaneko, H., 2001. Stoichiometric studies of
H2 generation reaction for H2O/Zn/Fe3O4 system. International Journal of Hydrogen Energy 26, 917922.
Trovarelli, A., 1996. Catalytic properties of ceria and CeO2containing materials. Catalysis Reviews in Science and
Engineering 38, 439520.
Tsakaloudi, V., Papazoglou, E., Zaspalis, V.T., 2004. Microwave ring of MnZn-ferrites. Materials Science and Engineering B 106 (3), 289294.
Weidenka, A., Nuesch, P., Wokaun, A., Reller, A., 1997.
Mechanistic studies of the water-splitting reaction for
producing solar hydrogen. Solid State Ionics 101103 (2),
915922.