Solar Energy 139 (2016) 676–694

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

Exploitation of thermochemical cycles based on solid oxide redox

systems for thermochemical storage of solar heat. Part 5: Testing of
porous ceramic honeycomb and foam cascades based on cobalt and
manganese oxides for hybrid sensible/thermochemical heat storage
Christos Agrafiotis ⇑, Andreas Becker, Martin Roeb, Christian Sattler
Deutsches Zentrum für Luft- und Raumfahrt/German Aerospace Center (DLR), Linder Höhe, 51147 Köln, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Cascaded ThermoChemical Storage (CTCS) of solar energy is a concept targeted to increase the volumetric
Received 21 April 2016 energy storage density and address the thermocline temperature distribution inside regenerative
Received in revised form 5 September 2016 sensible-only storage systems. CTCS involves the use of cascades consisting of different thermochemical
Accepted 15 September 2016
systems, distributed in a rational pattern inside the storage module tailored to their thermochemical
Available online 30 September 2016
characteristics and to the local heat transfer medium temperature. In the case of air-operated Solar
Thermal Power Plants such cascades can consist of porous structures incorporating different redox pair
Keywords:
oxide materials that can come in direct contact with the air stream used as heat transfer fluid and operate
Solar energy
Cascaded thermochemical heat storage
as compact, hybrid sensible-thermochemical storage modules in series.
Cobalt oxide Having previously identified the Co3O4/CoO and Mn2O3/Mn3O4 redox pairs as the most promising
Manganese oxide single-oxide systems for solar energy thermochemical storage, lab-scale (£ 25 mm), Co3O4- and
Cordierite Mn2O3-coated, porous cordierite honeycombs and foams were prepared and tested with respect to their
Ceramic honeycombs thermochemical characteristics in one- and two-oxides cascaded configurations employing redox oxide
Ceramic foams quantities in the range 15–150 g. For such Co3O4-loaded cascades thermochemical storage was clearly
demonstrated as heat uptake/release at constant temperature under proper testing conditions. Besides,
the additive effect of thermochemical on sensible storage within the same storage volume was visualized.
The operating conditions of cascades including both Co3O4 and Mn2O3 were dictated by the redox beha-
viour of the Mn2O3/Mn3O4 pair. Under proper conditions, such two-oxides-cascades could undergo cyclic
reduction-oxidation and operate in complementary temperature ranges during oxidation; however the
thermal effects of only the CoO oxidation reaction could be materialized into temperature rise of the
air stream exiting the cascade.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction
A major advantage of Solar Thermal Power Plants (STTPs) oper-
ating with Concentrated Solar Power (CSP) vs. other solar energy
technologies is that electricity production can be decoupled from
sunshine hours by storing the heat and converting it to electricity
when needed allowing thus a fully-dispatchable supply. Among
the various CSP technologies, solar towers offer the potential of
high temperatures and thus high thermodynamic conversion effi-
ciencies. In the other hand, the technology of volumetric ceramic
receivers – where concentrated radiation is absorbed inside the
volume of a porous structure that is used to transfer this heat to
⇑ Corresponding author.
E-mail address: christos.agrafiotis@dlr.de (C. Agrafiotis).
a fluid passing through it – provides for the efficient use of air as
heat transfer medium despite its low heat transfer coefficient. This
receiver technology has been developed on a solar tower since the
early 1990s over a series of projects culminating to the first-of-its-
kind, 1.5 MWel such CSP plant, Solar Tower Jülich (STJ) in Germany.
In STJ (Alexopoulos and Hoffschmidt, 2010) air at atmospheric
pressure flows through a solar-irradiated volumetric receiver made
of SiC honeycombs, being thus heated up to
700 °C to power a
steam generator. In parallel, a part of the solar-heated air is
diverted to a regenerative, sensible heat, Thermal Energy Storage
(TES) system (Tamme et al., 1991) where it transfers its enthalpy
to a solid medium as it flows through it (‘‘charging”). The storage
module consists of a 7 m  7 m  6 m housing (Fig. 1b) filled with
flow-through ceramic honeycombs to provide a large air-solid heat
exchange surface, according to the results of studies on

http://dx.doi.org/10.1016/j.solener.2016.09.013
0038-092X/Ó 2016 Elsevier Ltd. All rights reserved.
 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694 677

Fig. 1. (a) Layout of Solar Tower Jülich (STJ) power plant and (b) schematic of its sensible regenerative heat storage subsystem based on ceramic honeycombs (Zunft et al.,
2011); (c) example of air temperature distribution in such a regenerative atmospheric air storage system of an air-operated solar power plant (Fricker, 2004); (d) principle of
cascaded thermochemical storage configuration and operation during charging (left) and discharging (middle) of such a system, tailored to the characteristics of the Co3O4/
CoO and Mn2O3/Mn3O4 redox pairs and exemplary schematic (right) with such oxides coated/shaped in various porous supports with possibilities of porosity variation along
and across the cascade; (e) and (f) porous cordierite honeycombs and foams from various suppliers and of various sizes (<20 mm, £ 25 mm and £ 90 mm for testing in a TGA,
the current testing rig and a solar receiver respectively) before and after their coating with Mn2O3.

optimization of volumetric energy storage density (kWh/m3) and
pressure drop (Zunft et al., 2014, 2011). During off-sun operation
(‘‘discharging”) the air flow is reversed: ‘‘warm” air is introduced
through the lower end of the storage medium to be heated by that
as it flows towards the already ‘‘hotter” top end, before introduced
again to the power block.
In a regenerative sensible TES system, operation starts from an
isothermal cold or hot core, charging or discharging it with a
constant air mass flow rate of constant temperature. The heat trans-
fer from the air to the solid and vice versa must take place with a
small temperature difference, i.e. with a small exergy loss. The
result is a small temperature drop when discharging and a good
storage material utilization. Initially, during charging for example,
the temperature profile along the solid storage medium is stratified
as it is shown in Fig. 1c (Fricker, 2004). At first the layer of the solid
medium adjacent to the air inlet is heated to the constant, inlet air
678 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694

temperature; the other end of the solid remains practically at its
initial temperature, whereas in-between exists a thermocline zone
(TCZ) of the storage material where the heat transfer takes place
and the charging air is cooled. This ‘‘constant high-temperature
profile” moves progressively along the storage medium from the
‘‘hot” to the ‘‘cold” end, until finally the whole medium reaches a
temperature close to that of the incoming heat transfer fluid. A sim-
ilar procedure is applied for the discharging stage, where the dis-
charging air is heated in the TCZ. The longer the zones of
practically constant temperature above and below the TCZ are in
relation to the storage core flow path, the better is the utilization
of the storage material and the longer the time during which the
storage module can supply practically constant temperature.
Despite the relatively large volume of the storage module
(
120 m3) its’ full load discharge period is limited to
1.5 h. To
extend the storage capacity and achieve round-the-clock power
supply, rather than further augmenting the volume, the volumetric
energy storage density should be increased. As discussed in the
previous parts of this publication series, regenerative sensible heat
storage can be easily ‘‘hybridized” with TCS with redox oxides: in
addition to sensible-only-effects the heat effects of chemical reac-
tions can be exploited within the same storage module volume
increasing thus the volumetric energy storage density. In this per-
spective, the idea of exploiting redox oxide pairs coated on or
shaped into porous structures like honeycombs and foams, as
hybrid sensible-thermochemical solar energy storage means in
air-operated STTPs was set forth, having currently reached the
level of small-size redox-oxide coated honeycombs and foams
(Agrafiotis et al., 2015b,c; Tescari et al., 2014) as well as larger-
scale porous structures made entirely of Co3O4 (Karagiannakis
et al., 2014, 2016; Pagkoura et al., 2015, 2014). This concept of
exploiting multi-channeled porous ceramics to enhance heat
transfer between a heat transfer fluid and a storage medium has
been adopted later in similar ‘‘hybrid” sensible-latent heat storage
studies where bricks (de Gracia and Cabeza, 2015) or SiC honey-
combs (Li et al., 2015) have been combined with Phase Change
Materials (PCMs). In parallel, other TCS reactor concepts in CSP
plants are investigated: a recent simulation study (Ströhle et al.,
2016) has addressed TCS with packed and fluidized beds as storage
media using the Mn2O3/Mn3O4 pair as a model system; ‘‘moving
constant-temperature profiles” similar to the ones of Fig. 1c were
shown to describe the operation of such a storage module as well.
However, due to the thermocline temperature distribution
within the regenerative storage block, if the entire honeycomb
inventory was coated with a single oxide, the two TCS steps (charg-
ing and discharging) would cease to take place in regions where
and when the temperature would drop below the oxide’s reduc-
tion/oxidation temperature, respectively. To address such issues,
the concept of Cascaded ThermoChemical Storage (CTCS)
(Agrafiotis and Pitz-Paal, 2013) has been proposed where various
oxides can be cascaded with decreasing reduction temperature
from the ‘‘hot” to the ‘‘cold” end. Among four such single oxides
investigated by the authors in previous works (Agrafiotis et al.,
2015a, 2016a,b) only the Co3O4/CoO and Mn2O3/Mn3O4 pairs were
capable of reproducible, close-to-stoichiometric, cyclic operation.
Whilst the reaction schemes and thermodynamic calculations for
a particular redox pair assume one, common ‘‘equilibrium” tem-
perature for reduction and oxidation, this is not the case in prac-
tice: not only there might be some hysteresis phenomena during
operation that induce small differences of the order of few tens
of degrees between reduction and oxidation, but the Mn2O3/
Mn3O4 pair is characterized by a difference between reduction
and oxidation temperature of the order of 100–200 °C. Based on
the reduction/oxidation temperature ranges of these systems as
shown in the schemes (1) and (2) below, a two-oxides cascade,
each one coated or shaped to a honeycomb ceramic can be con-
ceived as per the sketch in Fig. 1c.
2 Co3 O4 þ DH ¢ 6CoO þ O2 Tred
890  C Tox
870  C ð1Þ
6Mn2 O3 þ DH ¢ 4Mn3 O4 þ O2 Tred
950  C Tox
690  780  C
ð2Þ
A drawback of this configuration is that during on-sun opera-
tion the entire cascade has to be heated up above the higher reduc-
tion temperature, that of Mn2O3; in the other hand during off-sun
operation the ‘‘warm” air entering from the bottom needs to initi-
ate first the Mn3O4 oxidation reaction taking place at much lower
temperature than the CoO oxidation.
The CTCS concept is independent of the reactor/heat exchanger
design and can be equally well applied to all possible heat exchan-
ger concepts like packed beds or honeycomb/foam-based ones and
to any combination of them. CTCS can be also combined with
porosity variation along and across the storage module: for exam-
ple honeycomb or foam structures with varying cross–section
porosity can be distributed along the cascade (rightmost sketch
of Fig. 1d) to homogenize and ‘‘smooth” the air velocity profile
achieving thus more uniform heat transfer. Furthermore it can be
combined with different (non-thermochemical) modes of solar
heat storage, like sensible or latent heat, in a configuration where
particular stages will employ for example only sensible and others
hybrid sensible-thermochemical or sensible-latent heat storage
(Zanganeh et al., 2015).
To further advance these concepts, the present work involved
the preparation of lab-scale (£ 25 mm), Co3O4- and Mn2O3-
coated, porous cordierite honeycombs and foams and their testing
in a one- or two-oxides cascaded configuration. Studies were per-
formed to identify the effects of operational parameters on oxygen
release/uptake and the thermal effects of the reactions on the air
stream exit temperature as well as to compare such hybrid sensi-
ble/thermochemical storage systems vs. respective sensible-only
ones.
2. Experimental
2.1. Materials
The same single oxides of Co3O4 and Mn2O3 from Materion
Advanced Chemicals as in Part IV of this series were employed.
Cordierite honeycombs of 400 cells per square inch/cpsi (Corning
S.A., U.S.A.) and foams of 20 and 30 pores per linear inch/ppi
(Ceramiques Techniques et Industrielles, France) as well as of 45
ppi (Fraunhofer-IKTS, Germany), all of 25.4 mm (1 in.) diameter
were used as porous supports. Their physical properties were pro-
vided in Part II of this work (Agrafiotis et al., 2015b); it was also
shown therein by XRD analysis after cyclic redox operation of
Co3O4-coated small-size cordierite specimens, that no formation
of mixed phases between cobalt oxide and cordierite has taken
place, justifying the current selection.
2.2. Coating of the ceramic structures
Series of such honeycombs and foams were coated with the two
powders via the approach known with the name ‘‘washcoating”.
This technique, described in details in previous publications
(Agrafiotis and Tsetsekou, 2000a,b) involves impregnation of the
supports in properly prepared low-viscosity, high-solids-content
powder slurries and subsequent drying and calcination. The details
for the preparation of the Co3O4 slurry were identical to those
already reported (Agrafiotis et al., 2015b). The same dispersant
 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694 679

(DOLAPIX PC75Ò, Zschimmer & Schwarz GmbH & Co KG Chemisc-
she Fabriken, Germany) was employed to stabilize the Mn2O3
slurry and shift its pH from its’ ‘‘natural” value of 4.7 which was
very close to its iso-electric point (IEP) reported in the literature
(Kosmulski, 2009), to
7.7. Indeed, low-viscosity slurry was
formed and no settling was observed during coating. After drying
for 12 h at 120 °C, the coated ceramics were calcined in a furnace
at 800 °C for 4 h heated with a rate of 2 °C/min to remove the
organic dispersant and induce adequate adhesion of the redox
coating on the supports. The temperature of 800 °C was selected
so that not to exceed the reduction temperature of either oxide.
The mass of oxide coated on the ceramic supports, calculated by
weighting the dry objects before coating and after calcination, is
reported in Table 1. A typical set of such supports spanning the size
range from very small ones suitable for the Thermogravimetric
Analyzer/TGA, to £ 25 mm ones suitable for the current test rig
and to £ 90 mm suitable for a solar receiver (Agrafiotis et al.,
2016a) before and after being coated with Mn2O3 are shown in
Fig. 1e and f respectively.
2.3. Experimental setup for sensible-thermochemical storage
experiments
The experimental evaluation of thermochemical/sensible heat
storage capacity was performed in a dedicated test rig consisting
of an alumina ceramic tube (65 cm long, inside diameter 27 mm)
placed inside a tubular furnace (Gero GmbH & Co. KG, Germany)
of maximum operating temperature 1250 °C programmed through
a Eurotherm 3200 programmer/controller. As shown in Fig. 2, the
heated part of the tube can accommodate cascades of non-coated
or redox-oxide-coated, 25 mm-diameter honeycombs or foams for
testing their sensible as well as thermochemical storage perfor-
mance or powder samples contained in ceramic boats. Both ends
of the reactor tube were closed with metal seals having ports for
gas inlet/outlet as well as for thermocouples that could be slided
in and out. The temperature at specific points along the reactor
was measured with S-type thermocouples placed at positions corre-
sponding as shown in the exemplary sketches of Fig. 2d. The oxygen
concentration at the reactor’s exit stream was monitored by an in-
line oxygen sensor (Sensor Technics Model XYA5M, measuring
accuracy ± 5 mbar, sampling frequency of 0.95 Hz). The air flow rate
was controlled by a mass flow controller (MKS INSTRUMENTS).
2.4. Experiments with oxides in powder form
At first Co3O4 and Mn2O3 were tested in powder form, sepa-
rately and in combination, placed in Al2O3 boats in a single- or
double-boat cascade configuration, as shown in the photo in Fig. 2c
and the sketches in Fig. 2d. All such experiments were performed
with an air flow rate of 80 sccm. The experimental protocols were
based on the results with these powders already reported in Parts
I–IV of this series. Since oxidation of Mn2O3 takes place with a

Table 1
Amount of redox powder coated on the various supports and respective loading percentages.

Co3O4 – coating
Code Weight After 1st calcination After 2nd calcination
Specimen Plain (g) Total weight (g) Total weight gain (g) Total weight gain (%) Total weight (g) Total weight gain (g) Total weight gain (%)
Cordierite honeycombs 400 cpsi, L = 50 mm, £ 25 mm (Corning)
1 11.85 18.98 7.13 60.2 34.61 22.76 192.1
2 12.40 23.18 10.78 86.9 38.70 26.30 212.1
3 11.86 18.77 6.91 58.3 38.25 26.39 222.5
4 11.70 18.94 7.24 61.9 33.84 22.14 189.2
Sum 32.06 97.59

Cordierite foams 20 ppi, L = 25 mm, £ 25 mm (CTI)

1 9.01 13.16 4.15 46.0 23.11 14.10 156.5
2 8.74 13.04 4.30 49.2 22.50 13.76 157.4
3 8.46 13.03 4.57 54.0 26.67 18.21 215.2
4 9.00 13.00 4.00 44.4 25.01 16.01 177.9
Sum 17.02 62.08

Cordierite foams 30 ppi, L = 25 mm, £ 25 mm (CTI)

1 10.13 16.27 6.14 60.6 30.05 19.92 196.7
2 11.43 18.16 6.73 58.9 31.95 20.52 179.5
3 11.59 17.12 5.53 47.7 28.44 16.85 145.4
4 10.51 16.10 5.59 53.2 27.65 17.14 163.1
Sum 23.99 74.43
Sum of 8 foams 41.01 136.51

Mn2O3 – coating
Cordierite honeycombs 400 cpsi, L = 50 mm, £ 25 mm, (Corning)
1 9.69 15.23 5.54 57.2 27.42 17.73 183.0
2 8.60 11.46 2.85 33.1 20.47 11.87 137.9
3 6.46 13.47 7.00 108.4 27.03 20.56 318.2
4 6.48 8.81 2.33 36.0 20.26 13.78 212.7
5 6.20 11.54 5.33 86.0 21.73 15.50 250.0
6 6.20 9.66 3.45 55.6 16.84 10.61 171.0

Cordierite foams 45 ppi, L = 50 mm, £ 25 mm, (IKTS)

1 4.99 8.84 3.85 77.2 18.58 13.59 272.5
2 5.22 9.29 4.06 77.8 21.34 16.11 308.5
3 6.12 10.50 4.39 71.8 23.73 17.61 288.0
4 4.97 9.59 4.62 92.9 19.93 14.96 300.9
680 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694

Fig. 2. Thermochemical/sensible storage capacity test rig: (a) furnace; (b) non-coated cordierite honeycombs ‘‘wrapped” with YSZ sheet before placed in the reactor; (c)
alumina reactor tube and representative configurations that were tested (from top to bottom): cascade of six non-coated honeycombs, of four redox-oxide–coated
honeycombs, of two boats containing loose redox oxide powder(s), of eight small-length foams before and after their coating with Co3O4 and of three redox-oxide–coated
honeycombs; (d) schematic of thermocouples placement in the cases of testing cascades of porous supports and one or two boats.

slower rate and at much lower temperatures than its reduction, to favorable for oxidation (Agrafiotis et al., 2016a,b). Experiments of
ensure completion of oxidation either long dwell times at suitable the first type have been reported previously (Agrafiotis et al.,
low temperatures or slow ramp rates had to be selected so that the 2015a); this work includes primarily experiments of the second
material remains for a long period within the temperature range type.
 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694
2.5. Storage experiments with redox-oxide-coated and non-coated
porous supports
Subsequently, experiments with non-coated and redox-oxide-
coated porous supports (honeycombs and foams) were performed
to compare sensible-only to hybrid sensible-thermochemical heat
storage under a variety of experimental conditions. For all the tests
involving porous specimens the supports were wrapped with a
thin sheet of Zirconium oxide, high-temperature felt insulation
(ZYF-100, Zircar Zirconia, Inc., U.S.A.) that acts simultaneously as
a flow sealant, as an insulating layer and prevents any interaction
between the redox-oxide-coated specimens and the alumina tube
(Fig. 2b). To exploit the whole heating length of the furnace for
the testing of honeycombs, six of them (50 mm-long each) were
placed in the reactor tube. For the tests of the foams, shorter
(25 mm-long each) 20 and 30 ppi structures were set together in
a cascade alternating one another with each of them used four
times (Fig. 2c). In all cases, the two honeycombs placed on either
end of the cascade were never coated but were employed to make
it possible to place thermocouples through their channels inside
the storage system (Fig. 2c and d).
The tests were performed with the same heating schedules as in
the prior tests with powders, however in this case a wide range of
air flow rates from 80 to 10,000 sccm was tested. As already men-
tioned, the temperature in the system must reach at least 950 °C
for both the oxidized states of the thermochemical storage oxide
materials to be reduced. However, in contrast to experiments in
a TGA, the possible reaction zone in the furnace configuration is
much longer and not all of it is subjected to the same temperature;
even when a fixed temperature set-point is programmed in the fur-
nace controller, a non-uniform temperature profile is established
along the reactor with higher temperatures near the center of
the heated zone and lower temperatures near both ends. In fact,
in these experiments the actual temperature reached in each hon-
eycomb depends on the air flow rate selected. For example, as dis-
cussed later in the text, high flow rates might cool a significant part
of the entrance length of the cascade to temperature levels below
those needed to induce reduction; the redox material in this region
will be practically non-available for thermochemical operation.
One of the goals of this experimental series was therefore to deter-
mine the effect of air flow rate on the temperature distribution of
the cascade along the reactor in comparison to the reduction tem-
perature of each particular system. Thereafter, experiments were
performed with the hybrid sensible-TCS setup with the honey-
combs and foams after their first as well as their last coating at
specific air flow rates to investigate the effect of redox material
employed. Furthermore, the system was tested with an empty
reactor tube. These tests can be viewed as blank tests under ‘‘no
storage conditions”, either sensible or thermochemical, for com-
parison. Finally, based on the results of the tests above, experi-
ments comparing Co3O4-only, Mn2O3-only and mixed honeycomb
cascades were performed.
3. Results – discussion
3.1. Experiments with Co3O4 and Mn2O3 oxides in powder form
The first sets of experiments were performed with boats con-
taining only Co3O4, based on its redox behaviour when tested in
the TGA, where complete reduction-oxidation was observed during
cycling between 985 and 785 °C irrespective of the ramp rate. Such
an experiment was performed in the current test rig with a boat
containing 3 g of Co3O4 under a ramp rate of 5 °C/min. The oxygen
traces during test (black curves in Fig. 3a) demonstrate cyclic
reduction-oxidation of Co3O4 and reproducible oxygen produc-
681
tion/consumption curves from cycle to cycle. In contrast to the
respective TGA experiments where the weight change of the sam-
ple is recorded within a very short time, oxygen detection takes
longer and under the particular ramp rate the transition of the oxy-
gen concentration from reduction to oxidation looks ‘‘quasi-
continuous” and not characterized by a plateau of oxygen concen-
tration at
21% as is the transition from oxidation to reduction due
to the higher temperature range spanned in the latter case.
Next experiments involved separate tests with Co3O4 and
Mn2O3 powders under temperature schedules to ensure reduction
of both powders. The first such experiments involved one heating
cycle with 5 °C/min to 1000 °C to exceed the reduction tempera-
ture of both oxides and a dwell of 2 h at this temperature to ensure
completion of reduction, before letting then the system cool down
with the ‘‘natural” rate. For the case of Co3O4 the temperature and
oxygen concentration evolution curves during two such experi-
ments with quantities of 8 and 10 g are also shown in Fig. 3a, to
delineate the effect of Co3O4 mass on the peak height and area.
The comparative performances of the Co3O4 and Mn2O3 powders
under exactly the same temperature schedule are shown in Fig. 3b.
Whereas the reduction of the two powders takes place in the same
temperature range and the behaviour of Co3O4 is fully reversible
upon cooling, the ‘‘natural” cooldown rate (
5 °C/min) is too high
for Mn3O4 oxidation to take place and no oxygen is recorded to be
uptaken from the reduced powder. Other than that, the shape of
the two oxygen release curves is similar, characterized by a sharp
initial increase followed by a parabolic-like drop, similar to that of
such oxygen release curves of redox materials used for water/car-
bon dioxide splitting obtained during thermal reduction (Agrafiotis
et al., 2013; Kostoglou et al., 2014).
In an effort to induce oxidation of Mn3O4 under cycling condi-
tions, a slower ramp rate of 2 °C/min was tested, within 4 temper-
ature cycles between 1090 and 640 °C and without dwell between
the steps. The last cooling step was performed with 2 °C/min until
640 °C and with the ‘‘natural” cool-down rate thereafter until
ambient temperature. The results are shown in Fig. 4a (full
4-cycle experiments) and b (‘‘magnification” of the first cycle).
The absolute value of the O2 curve of the Mn2O3-only experiment
corresponds to the scale on the right axis; the other two O2 curves
have been ‘‘shifted” lower for reasons of clarity. Since the oxygen
absorbed/released per g of Co3O4 is much higher than that per g
of Mn2O3, different quantities of each material were employed
(as mentioned on the graphs) to ‘‘produce” peaks of approximately
the same height. Under these conditions reduction and oxidation
take place for both materials with a clear, repeatable way both in
their single tests as well as when they are together. The reduction
peaks for both materials start to appear upon heating, at around
900–920 °C. During cool-down, oxidation of CoO takes place first,
around 900 °C. However, any oxygen uptake due to oxidation of
Mn3O4 is not recorded during further cool-down, but only when
the system is heated up again, around 630 °C (Fig. 4b), lasting until
a temperature of
830 °C is reached. This observation is in accor-
dance with relevant TGA experiments under a similar schedule
reported in Part IV of the series (Agrafiotis et al., 2016b) where
the major weight gain of the reduced Mn3O4 powder due to re-
oxidation was recorded upon heat-up from the lower temperature
limit.
3.2. Hybrid sensible-thermochemical storage experiments with Co3O4-
coated porous ceramic supports
The first such experiments were exploratory targeted on identi-
fying conditions (with respect to air flow rate, amount of coated
material etc.) where the thermochemical reactions could induce
measurable effects on the oxygen concentration and, most impor-
tantly, on the temperature of the air stream flowing through the
682 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694

1200 40
Co3O4

T=990oC Air flow rate = 80 sccm; Ramp rate:5oC/min

1000 mCo O = 3g
3 4

mCo O = 8g
30
3 4

O2 concentration (% in air)
mCo O = 10g
3 4

T=790oC Temperature
800 O2 concentration
Temperature (oC)

600 20

400

10

200

0 0
(a) 0 200 400 600 800 1000
Time (min)

1200 40
Air flow rate = 80 sccm
Tplateau: 1000oC; Dwell time=2 hrs
Heating rate:5oC/min
1000 mCo O = 8 g; mMn O = 9 g
3 4 2 3

O2 concentration (% in air)
30
Temperature (oC)

mOxygen : ~ 70 %
800
of theoretical

600 20

mOxygen : ~ 71 %
Temperature
of theoretical
400 Co3O4
Mn2O3

mOxygen : ~ 86 % O2 concentration 10
Co3O4
of theoretical
200 Mn2O3

(b) 0 0
0 200 400 600 800 1000
Time (min)

Fig. 3. Temperature and O2 concentration evolution curves during experiments with boats containing only Co3O4 or only Mn2O3 powder under an air flow rate of 80 sccm and
heating rate of 5 °C/min: (a) experiments with Co3O4, either cycled between 990 and 790 °C without dwell (black curves) or heated up to 1000 °C, dwell for 2 h and ‘‘natural”
cooldown (blue and red curves); (b) comparison between Co3O4 and Mn2O3 under the latter heating schedule and calculation of conversion extent of thermochemical
reactions from integration of respective O2 concentration curves. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version
of this article.)

coated supports. Due to the higher intensity of thermochemical
effects of Co3O4 vs. that of Mn2O3, all such experiments were per-
formed with Co3O4–coated supports: with cascades of 4 honey-
combs (50 mm-long each) and 8 foams (25 mm-long each)
coated only once and both having a total length of 20 cm as per
the photographs in Fig. 2c. This set of experiments was performed
under the first schedule above employed with powders, i.e. heating
with 5 °C/min up to a programmed temperature set point of
1070 °C, dwell for 2 h and then ‘‘naturally” cool-down to ambient
temperature. Low air flow rates, between 30 and 370 sccm were
selected at first to better visualize any change in oxygen concentra-
tion in the air stream due to the thermochemical reactions taking
place.
Fig. 5a shows the results with respect to O2 concentration and
temperature at the end of the reaction zone (i.e. at the end of the
last coated support) for the cascade of 8 foams after their first
Co3O4–coating under two air flow rates, 180 and 280 sccm. The
two oxygen curves exhibit nearly the same profile during reduc-
tion and oxidation. At higher flow rates the peaks become smaller
since the same amount of oxygen is produced/consumed per g of
redox material but this quantity has a smaller effect on the total
concentration of oxygen in the air stream. The heat effect of reduc-
tion can be distinguished qualitatively by the small ‘‘deflections” in
the temperature profile during heat-up: during the time when
reduction takes place the increase of temperature is delayed since
heat is consumed in the endothermic reduction reaction and not to
sensibly heat the gas. As soon as reduction is complete the air
stream temperature rises again. However during cool-down, the
heat generated by the oxidation reaction is not enough to induce
a respective clearly visible change in the temperature curve for nei-
ther of the two flow rates. It should be noted that the apparently
low ‘‘reduction” temperature (
770 °C) where abrupt oxygen con-
centration change is recorded (in this as well as in all other subse-
quent experiments in a similar configuration) is misleading: this
 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694 683

Fig. 4. Temperature and O2 concentration evolution curves during experiments with the same configuration as in Fig. 3, under a schedule of 4 temperature cycles between
1090 and 640 °C, with a ramp rate of 2 °C/min, without dwell between the steps and ‘‘natural” final cooldown: (a) entire experiments; (b) ‘‘magnification” of the first cycle.

temperature is measured at the end of the reaction zone whereas
the oxygen concentration change is due to reduction of redox
material being already at higher temperatures within the ‘‘central”
areas of the reaction zone. Fig. 5b shows the respective results for
the case of a cascade of 4 honeycombs after their first Co3O4–coat-
ing for five air flow rates between 30 and 370 sccm. The height and
area of the oxygen peak during reduction is again inversely propor-
tional to the air flow rate. It is interesting to note that for the par-
ticular mass of Co3O4 coated, air flow rates lower than 180 sccm do
not provide enough oxygen for the oxidation reaction to fully take
place and therefore during cool-down oxygen concentration drops
to zero: all oxygen available in the flowing air stream is consumed
by the reduced oxide CoO but still is not enough to oxidize it com-
pletely. As a result, since the experiments took place consecutively
without taking the honeycombs out of the furnace but by simply
changing the flow rate from experiment to experiment, an oxygen
peak downwards, corresponding to oxidation of the reduced oxide
that was not completely oxidized in the previous experiment just
before the peak upwards corresponding to reduction, is visible in
several experiments. Just like in the case of foams, the heat effect
of reduction is distinguished by the small ‘‘bend” in the tempera-
ture profile during heat-up, but any temperature increase due to
oxidation taking place during cool-down is very low and not
detectable in any of the temperature curves during cool-down that
practically coincide.
In all previous experiments with the Co3O4-coated structures
the oxygen concentration effects during oxidation were clearly
depicted; however any respective temperature effects from the
enthalpy of the oxidation reaction could not be visualized as a
recordable increase of the air stream exit temperature. Even
though the coated material reacts, the heat produced by its oxida-
tion is not transferred to the flowing air stream but is obviously
dissipated and lost through the reactor walls. To increase the pos-
sibility of visualizing the heat effects during oxidation, the honey-
combs and foams were further coated with Co3O4 with consecutive
impregnations in the slurry to a much higher final total loading (to
generate more heat per unit volume). Subsequently similar exper-
iments were performed but employing much higher air flow rates
684 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694

Fig. 5. Oxygen concentration evolution and temperature at the end of the reaction zone, i.e. at the end of the last coated support, during experiments under heating with 5 °C/
min up to 1000 °C, dwell for 2.5 h and ‘‘natural” cool-down to ambient temperature: (a) cascade of 8 foams after their first Co3O4–coating under air flow rates of 180 and 280
sccm; (b) cascade of 4 honeycombs after their first Co3O4–coating for five air flow rates between 30 and 370 sccm.

to enhance the transport of the heat produced by the reaction
enthalpy by convection along the reactor towards the exit of the
cascade, before this heat is dissipated.
The results of these experiments for the two systems, honey-
combs and foams, are shown in Figs. 6 and 7 respectively. For both
kinds of supports, the effects of increased coated mass (that has
been approximately tripled) and increased air flow rate and conse-
quently heat transfer by convection, can be clearly observed on the
temperature profiles comparing Figs. 6a and 7a vs. Fig. 5. At lower
flow rates the plateau temperature reached during dwell is practi-
cally the same for all flow rates. However beyond a threshold value
of air flow rate the plateau temperature achieved during dwell
increases. In other words, higher flow rates enhance heat transfer
by convection and thus the point at the end of the reaction zone
reaches higher temperatures (for the same furnace temperature
set-point programmed). In addition, the heat effects of the chemi-
cal reactions become evident from the ‘‘magnifications” of the tem-
perature profiles shown in Figs. 6b and 7b: at higher flow rates the
characteristic ‘‘deflections” during reduction become much more
pronounced and a short, but clearly visible ‘‘quasi-isothermal”
range (‘‘quasi-horizontal” line segment) appears where the tem-
perature remains practically constant for as long as reaction takes
place. Upon reaction completion the air stream temperature rises
again abruptly to follow the ‘‘straight” line of linear increase with
time (exemplary dotted blue line segment). Furthermore, such
quasi-isothermal segments appear now clearly also during the
duration of oxidation in the cooldown step. With increased quan-
tity of redox material within the same volume and enhanced con-
vection the ‘‘extra” heat produced by oxidation can be transferred
more effectively downstream the reaction zone. These phenomena
were common to both foams and honeycombs. In conclusion, the
respective ‘‘plateau” segments of the air stream exit temperature
demonstrate qualitatively the effects of chemical reaction
enthalpies and the principle of thermochemical storage operation,
 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694 685

Fig. 6. Experiments with a cascade of 8 foams after their final Co3O4–coating, under the same schedule as those in Fig. 5: (a) effects of air flow rate (range 180–5000 sccm) on
oxygen concentration evolution and temperature at the end of the reaction zone; (b) ‘‘magnification” of temperature profiles to delineate the thermal effects of chemical
reactions.

i.e. heat storage/release under constant temperature. These results
are in full qualitative agreement with results from a very recent
study with similar honeycomb structures, either extruded from
Co3O4-based compositions or Co3O4-coated cordierite ones
(Karagiannakis et al., 2016). The differences are due to the slightly
different experimental testing configuration employed. The
authors employed per experiment only one redox oxide honey-
comb of length between 23 and 50 mm and of redox powder
weight between 16 and 31 g. Thus the whole honeycomb was prac-
tically at the same temperature irrespective of the flow rate, but
the amount of redox material and consequently of the enthalpy
released/absorbed is small. Therefore, in their study the thermo-
chemical operation of Co3O4 was manifested as a slower increase
of the temperature at the exit of the honeycomb when oxygen
was produced during heating and as an air exit temperature higher
than that of entrance when oxygen was uptaken during cooling. In
our study a much higher amount of redox material is employed but
the length of the redox zone is not under the same temperature
thus the thermochemical operation of Co3O4 cannot be manifested
as an air exit temperature higher than that of entrance during oxi-
dation; instead the thermochemical effects are manifested as a
constant temperature plateau at the exit of the reaction zone dur-
ing both steps.
3.3. Sensible-only storage experiments with non-coated porous
ceramic supports
Based on the discussion above, a further issue of concern rele-
vant to the particular cascade configuration of coated porous spec-
imens had to do with whether all the quantity of the redox oxide
loaded on the honeycombs was above the reduction/oxidation
temperature and was therefore available for thermochemical reac-
tion. To clarify this issue, experiments in sensible-only storage
mode (i.e. with non-coated honeycombs) were performed varying
the flow rate and measuring the temperature along the reactor, to
determine what extend of the cascade is above the reduction tem-
peratures of the two oxides examined herein, Mn2O3 and Co3O4,
950 °C and 890 °C respectively. Characteristic temperatures at four
686 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694

(a) 1200 60
4 Coated honeycombs, higher flow rates, higher loadings

Temperature at reaction zone end

Temperature at reaction zone end (C°)

5000 sccm
1000 2500 sccm 50
1000 sccm

O2 concentration (% in air)
370 sccm
Total Co3O4 coated mass = 98 g
O2 concentration
800 5000 sccm 40
2500 sccm
1000 sccm
370 sccm

600 30

400 20

200 10

0 0
0 200 400 600 800 1000
Time (min)

(b) 100
4 Coated honeycombs, higher flow rates, higher loading, total Co3O4 coated mass = 98 g

1100
Temperature at reaction zone end (C°)

O2 concentration (% in air)
75

1000

50
900

25
800

700 0
150 200 250 300 350 400 450
Time (min)

Fig. 7. Experiments with a cascade of 4 honeycombs after their final Co3O4–coating, under the same schedule as those in Fig. 5: (a) effects of air flow rate (range 370–5000
sccm) on oxygen concentration evolution and temperature at the end of the reaction zone; (b) ‘‘magnification” of temperature profiles to delineate the thermal effects of
chemical reactions.

different points along the cascade of six non-coated cordierite hon-  For even higher flow rates, 5000, 7500 and 10,000 sccm even
eycombs as per the sketch in Fig. 8a are shown in Fig. 8b compared the 3rd honeycomb becomes exposed (at least partially) at tem-
to these two reduction temperatures. The trends of the green and peratures lower than the reduction temperatures of both oxides
blue curves point out that increasing the air flow rate ‘‘cools” the and therefore not entirely available for a cascaded TCS concept
entrance of the cascade. The trends of the red and orange curves in this particular rig.
in the other hand, demonstrate that higher air flow rates enhance
From the respective experiments with an empty tube shown in
the transfer of heat produced both resistively by the furnace as
Fig. 8c the effect of sensible-only storage vs. no-storage becomes
well as due to the enthalpy of thermochemical reactions, towards
evident: for the same flow rate, a higher temperature is achieved
the end of the cascade. The following conclusions can be drawn for
both near the center as well as near the exit of the cascade when
the particular test rig configuration from this graph:
the tube is filled with porous honeycombs. This effect is better
delineated in the graph of Fig. 9 where the experiments at air flow
 For flow rates 280 and 380 sccm, 4 honeycombs - the 2nd, 3rd,
rate of 5000 sccm with the four coated honeycombs (hybrid
4th and 5th - are entirely above the reduction temperature of
sensible-thermochemical storage), the non-coated honeycombs
Co3O4 (890 °C) and marginally above that of Mn2O3 (950 °C).
(sensible-only storage) and the empty tube (no storage) are com-
 For higher flow rates1000 and 2500 sccm, three honeycombs –
pared. Even though the first of the four coated honeycombs is
the 3rd, 4th and 5th - are above the reduction temperature of
not entirely above the reduction temperature of Co3O4, the heat
Mn2O3.
effects of the reduction reaction during heat-up and of the
 Flow rates of 1000 and 2500 sccm seem to be the ‘‘optimal”,
oxidation reaction during cool-down are clearly qualitatively
where the temperatures at the entrance of the 3rd and at the
reflected on the temperature of the air stream exiting the reaction
exit of the 5th honeycomb are practically equal.
zone. Cooling of this stream is indeed retarded due to the
 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694 687

T=1170oC
T5 T2 T1

Air flow Air flow

out T6 in

(a) 14.5 cm 2.5 cm 5 cm 5 cm 5 cm 5 cm 5 cm 5 cm 3 cm 15 cm

1400
6 non-coated cordierite honeycombs 400 cpsi
End of 1st honeycomb; End of 2nd honeycomb; End of 5th honeycomb; End of 6th honeycomb

1200 80
sccm
180
sccm
280
sccm
380
sccm
500
sccm
1000
sccm
2500
sccm
5000
sccm
7500
sccm
10000
sccm

1000 Tred Mn O = 950oC

2 3

Tred Co O = 890oC
Temnperature (oC)

3 4

800

600

400

200

(b) 0
0 2000 4000 6000 8000 10000 12000
Time (min)

1400
Empty tube th
"End of 5 honeycomb";
th
"End of 6 honeycomb";
st
"End of 1 honeycomb"

1200 80 sccm 180 sccm 380 sccm 1000 sccm 2500 sccm 5000 sccm

T = 1045oC T = 1045 C
o
T = 1030oC T = 1020oC
1000 T = 992oC T = 990oC

T = 923oC
Temnperature (oC)

o
T = 895oC T = 900oC o
T = 906 C
T = 875 C

T = 860oC T = 830oC
800 o
T = 795 C

T = 750oC

600 T = 553oC

T = 384oC
400

200

(c) 0
0 1000 2000 3000 4000 5000 6000
Time (min)

Fig. 8. Characteristic temperature profiles at four different points along the reactor in experiments under the same schedule as those in Figs. 5–7: (a) schematic of
arrangement and measurement points; (b) experiments with a cascade of 6 non-coated cordierite honeycombs; (c) experiments with an empty tube.

thermochemical reaction, demonstrating thus the ‘‘proof-of- ule as those depicted in Figs. 5–7, i.e. involved heating with 5 °C/
concept” of the idea of TCS. min up to a programmed temperature set point of 1170 °C and
dwell for 2 h, but cool-down to ambient temperature was per-
3.4. Hybrid sensible-thermochemical storage experiments with Co3O4- formed with a slow, controllable rate of 2 °C/min (similar to that
and Mn2O3-coated porous ceramic supports in the experiments with powders depicted in Fig. 4) instead of nat-
ural cooling, to investigate whether this would have any favorable
The next set of experiments involved comparison among effect on the oxidation of Mn3O4.
Co3O4-, Mn2O3-coated and mixed honeycomb cascades at the The temperature and oxygen concentration evolution curves as
higher range of flow rates already tested. Based on the above, in a function of air flow rate are shown in Fig. 10a for the Co3O4- and
these experiments in one hand only three instead of four redox- 10b for the Mn2O3-loaded honeycombs. The peak temperatures
coated honeycombs were employed, as per the last photograph measured are listed in the graphs. Small discrepancies between
in Fig. 2c. The first experiments were according to the same sched- temperatures at the same location between experiments with
688
1200
Temperature at reaction zone end (oC)
1000
800
storage effect
600
400
200
0 0
0 200
Fig. 9. Comparison of temperature at the end of the reaction zone among experiments with a cascade of 4 Co3O4–coated honeycombs (hybrid sensible-thermochemical
storage), non-coated honeycombs (sensible-only storage) and empty tube (no storage) under the same heating-cooling schedule as in Figs. 5–7 and air flow rate of 5000 sccm.
Co3O4- and Mn2O3-loaded honeycombs under the same air flow
rate are due to slight shifts of the thermocouples location. The tem-
perature trends are in full accordance with those depicted in Fig. 8;
increase of air flow rate renders temperature effects more visible
and oxygen concentration ones less pronounced. However, even
though the Oxygen traces of the Co3O4-loaded honeycombs are
clear both during heat-up and cool-down, oxygen peaks from the
Mn2O3-loaded ones are visible only during reduction (heat-up),
even at the lowest flow rate tested where the intensity of any such
peaks was expected to be higher. The ‘‘magnification” of the tem-
perature profiles at the exit of the reaction zone for the two sys-
tems under the same flow rate of 2500 (Fig. 10c) shows not only
the effects of Co3O4 reduction–oxidation on the ‘‘deflection” of
the temperature curves in complete analogy with Figs. 6b and 7b,
but also the absence of such effects in the case of Mn2O3. It is clear
therefore that even this low cool-down rate does not suffice to
induce oxidation of Mn3O4 back to Mn2O3; this is also reflected
on the appearance of oxygen peaks downwards during initial
heat-up in the experiments in Fig. 10b (which were actually per-
formed in a sequence opposite than that shown in the Figure, i.e.
from the highest to the lowest flow rate), corresponding to oxida-
tion of the material that was not oxidized during cool-down in the
previous experiment.
Thus the Mn2O3-loaded honeycombs were also further loaded
with Mn2O3 to their final loadings listed in Table 1 in order to
increase the heat per unit volume produced and a final set of com-
parative experiments among Co3O4-, Mn2O3-coated and mixed
honeycomb cascades was performed at the higher range of flow
rates, but this time in cyclic mode: under the same schedule of four
(4) cycles between upper and lower temperature set points of 1070
and 670 °C respectively, with a ramp rate of 2 °C/min and no dwell.
The respective curves of temperature and oxygen concentration
evolution are shown in Fig. 11 (in separate plots for reasons of clar-
ity) for three characteristic cases: 3 Co3O4-coated honeycombs
(Fig. 11a), 3 Mn2O3-coated honeycombs (Fig. 11b) and a cascade
of 1 Co3O4- and 2 Mn2O3-coated honeycombs (Fig. 11c). Just like
in the previous experiments, concentration effects are enhanced
in lower air flow rates and temperature at higher ones. Increasing
the flow rate to 5000 sccm results in temperatures lower than that
of the reduction temperature of Mn2O3 even at the first coated
honeycomb; for this reason in the experiment with 1 Co3O4- and
2 Mn2O3-coated honeycombs the Co3O4-coated one was placed
C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694
50
4 Co3O4-coated honeycombs
Total Co3O4 coated mass = 98 g
Air flow rate = 5000 sccm
Temperature at reaction zone end:
Coated honeycombs 40
Non-coated honeycombs
Empty tube
O2 concentration (% in air)
O2 concentration
Hybrid 30
Sensible-TC
Sensible-only
Storage effect
20
10
400 600 800 1000
Time (min)
‘‘first” at the entrance of the reaction zone, to ensure that the
Mn2O3-coated ones will be exposed at higher temperatures.
The effect of increasing flow rate on the oxygen concentration
trace is also evident. At higher flow rates the peak height dimin-
ishes and the relevant curves, even though maintaining their qual-
itative characteristics, are characterized by a lot of ‘‘noise”. The
oxygen evolution curves are compared for the case of the lower
flow rate tested, 500 sccm, where they are most intense, in
Fig. 12. The curve of the Co3O4-only experiment corresponds to
the absolute scale on the right axis and the other two have been
‘‘shifted” to lower absolute values. In this cycling mode the oxida-
tion reaction of Mn3O4 is evident, though not during the cool-down
step but during the next heat-up one, prior to reduction. In com-
plete analogy to the experiments with powders depicted in
Fig. 4, the curve of the ‘‘mixed” Co3O4-Mn2O3 cascade is a superpo-
sition of the other two, characterized by overlapping reductions
and two distinct oxidations, of CoO during cool-down and of
Mn3O4 essentially during heat-up.
To delineate the effects of redox oxide material and air flow rate
on temperature, the temperature profiles at the entrance and exit
of the reaction zone corresponding to the first cycle of the three
experiments depicted in Fig. 11, are shown in ‘‘magnification” in
Fig. 13a and b for the two cases of low (500 sccm) and higher
(2500 sccm) air flow rates respectively. Even though the effects
are of relatively low intensity, are at the same time qualitatively
visible and distinct. It is evident that the intensity of any deflection
of the temperature curves is ‘‘proportional” to the amount of Co3O4
present: during heat-up such deflection is obvious in the case of
Co3O4-only-coated honeycombs, present but less intense in the
‘‘mixed” Co3O4-Mn2O3 cascade and inconspicuous in the case of
Mn2O3-only cascade. It is interesting to note the effect of the Peclet
number on the temperature profiles: in the case of low flow rates
where thermal diffusion is significant, the temperature profiles at
the entrance of the reaction zone (green curves) exhibit a deflec-
tion as well, whereas at higher flow rates convection limits this
effect to the end of the reaction zone (red curves). During cool-
down only in the case of high flow rate and Co3O4-only coating, a
minute but nevertheless clearly present deflection can be observed
in the temperature at the exit of the cascade, shifting it to slightly
higher values (in the dotted box in Fig. 13b), in analogy to Fig. 9.
To summarize, even though the sequential chemical reaction
part of a Co3O4-Mn2O3 cascade during oxidation was
 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694 689

Fig. 10. Temperature and oxygen concentration evolution profiles during experiments under heating with 5 °C/min up to a set-point of 1170 °C, dwell for 2.5 h and cool-
down with a rate of 2 °C/min to ambient temperature: (a) 3 Co3O4-coated honeycombs (mass of Co3O4 = 75 g); (b) 3 Mn2O3-coated honeycombs (mass of Mn2O3 = 15 g); (c)
‘‘magnification” of temperature profiles at reaction zone end for the experiments with air flow rate 2500 sccm.

demonstrated, it was not possible to materialize the thermal effect der grade and quantity and in the specific test rig configuration
of the Mn3O4 oxidation reaction into quantitative heat transfer employed in this study, in contrast to that of CoO. Even though
fluid temperature changes, at least with the particular Mn2O3 pow- the performance of the Mn2O3/Mn3O4 system and especially its
690 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694

Fig. 11. Temperature and oxygen concentration evolution profiles during testing for four cycles between upper and lower temperature set points of 1070 and 670 °C
respectively, with a ramp rate of 2 °C/min and no dwell for: (a) 3 Co3O4-coated honeycombs (mass of Co3O4 = 75 g); (b) 3 Mn2O3-coated honeycombs (mass of Mn2O3 = 50 g);
(c) a cascade of 2 Co3O4- and 1 Mn2O3-coated honeycombs.

oxidation reaction was found to depend on the physical/morpho- above not only in studies involving TGA experiments (Carrillo
logical properties of the particular powder source (Carrillo et al., et al., 2014, 2015) but also in larger-scale packed powder bed test-
2014, 2015), the inherent peculiarities of this system mentioned ing rigs (Wokon et al., 2014) constitute a challenge. Obviously
 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694
Fig. 12. Comparison of oxygen concentration evolution profiles among the experiments of Fig. 10 under an air flow rate of 500 sccm (temperature profiles from experiment
with 3 Co3O4-coated honeycombs).
incorporation of this system in large-scale STPP operation needs
further elaboration of the operation strategy, especially since dur-
ing discharging, the oxidation reactions of such redox-oxide-based
thermochemical systems should occur upon their cooling (Fricker,
2004; Ströhle et al., 2016). For both systems, larger-scale, effi-
ciently insulated, multiple-honeycomb/foam systems with high
thermal mass can perhaps allow full exploitation of any thermo-
chemical benefits. These issues are currently under consideration
in next experimental campaigns targeted to testing of such cas-
caded systems in larger scale.
3.5. Oxygen production/consumption
In all such experiments as above, from the known quantities of
powder mass and air flow rate, the oxygen produced/consumed
per mass of powder (e.g. in lmoles O2/g of powder) with respect
to the theoretical one expected from a given mass of redox powder
according to fully stoichiometric reduction/oxidation reactions,
can be calculated by integrating the respective oxygen concentra-
tion curves. In the first set of experiments with powders (Fig. 3),
for both redox pairs Co3O4/CoO and Mn2O3/Mn3O4, this amount
of O2 calculated from such integrations of the curves vs. the max-
imum one corresponding to stoichiometric reduction/oxidation of
1 g of Co3O4 (2076 lmoles of O2) or Mn2O3 (1056 lmoles of O2)
during reduction ranged between 70% and 77% whereas that for
CoO oxidation accounted for 84–86% of the maximum value corre-
sponding to stoichiometric reaction (Fig. 3b and Table 2, no reoxi-
dation of Mn3O4 demonstrated).
In the next set of experiments with 8 Co3O4-coated foams and 4
Co3O4-coated honeycombs heated with a constant ramp rate of
5 °C/min to 1070 °C, dwelling for 2 h and then cooled-down to
ambient temperature under different flow rates (Figs. 5–7) calcula-
tion of the relevant percentages with respect to the theoretical
ones from the known amount of Co3O4 loaded on the honey-
combs/foams and integration of oxygen curves like in the case of
the experiments with powders, was problematic. At the lower
range of air flow rates (30–370 sccm) as already mentioned, the
experiments took place consecutively without taking the honey-
combs out of the furnace rig but by simply changing the flow rate
from experiment to experiment. In several of these experiments it
691
is obvious that the area under the oxidation oxygen peak during
cooling is higher than that under the sharp reduction peak during
the previous heat-up. What has really happened is that due to
shortage of oxygen during cooling in a previous experiment with
another flow rate, a quantity of the oxide has remained in its
reduced state; when the next experiment with a different air flow
rate started, at first during initial heatup an oxidation of this quan-
tity of the oxide was observed as an oxygen uptake peak down-
wards. However due to the fast heatup, not all the quantity of
this ‘‘leftover” reduced oxide has been oxidized before the reduc-
tion temperature is reached. Thus the quantity of oxygen evolving
beyond this stage corresponds to a mass of oxide being reduced
less than the stoichiometric one (since some quantity of the oxide
has been already remained reduced from the previous experi-
ment). Then during cooldown and oxidation the mass of oxygen
uptaken is not only due to the oxidation of the reduced oxide dur-
ing heat-up but in addition due to the oxidation of the reduced
oxide that is ‘‘carried” through the previous experiment, as a result
the area is higher than that from the peak upwards. This trend is
eliminated at much higher flow rates, where the areas under the
reduction curves are consistently higher than those under the oxi-
dation curve. However another problem was introduced: with air
flow rates above a threshold value the coated supports in the
entrance of the cascade were cooled below the reduction temper-
ature of Co3O4 and thus the powder loaded in them was not avail-
able for reduction. Since there was no way of knowing the mass of
Co3O4 actually participating in the reaction so that to calculate the
theoretical number of (moles of O2/g of Co3O4) corresponding to
‘‘stoichiometric” reduction/oxidation, calculation of a percentage
with respect to the theoretical maximum was not possible either.
In contrast, though, based on the above, conversion calculations
are not meaningless for several of the experiments of the 2nd type
depicted in Fig. 11, i.e. testing of 3 honeycombs, for four cycles
between 1070 and 670 °C with a ramp rate of 2 °C/min and no
dwell: in particular for these experiments where temperature
measurements show that all three oxide-coated honeycombs (with
a known total mass of redox oxide) are above the reduction tem-
perature (dotted red line) throughout the experiment. The most
characteristic such experiments are these with 3 Co3O4-coated
honeycombs (with a known total mass of Co3O4 = 75 g) under air
692 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694

1150
o
3 Co3O4- vs. 3 Mn2O3- vs. 2 Mn2O3+1 Co3O4 honeycombs, 2 C/min , 500 sccm
nd th
Co3O4: T end of 2 , T end of 5
1100 Co3O4+ Mn2O3: T end of 2 ,
nd
T end of 5
th

nd th
Mn2O3: T end of 2 , T end of 5

1050

Temperature ( C)
o

1000

o
Tred Mn O = 950 C
950
2 3

900

850

(a) 800
350 400 450 500 550 600 650 700 750
Time (min)

1150
o
3 Co3O4- vs. 3 Mn2O3- vs. 2 Mn2O3+1 Co3O4 honeycombs, 2 C/min, 2500 sccm

nd th
Co3O4: T end of 2 , T end of 5
1100 Co3O4+ Mn2O3:
nd
T end of 2 , T end of 5
th

nd th
Mn2O3: T end of 2 , T end of 5

1050
Temperature ( C)

1000
o

o
Tred Mn O = 950 C
950
2 3

900

850

(b)
800
5350 5400 5450 5500 5550 5600 5650 5700 5750
Time (min)

Fig. 13. Comparison of temperature evolution profiles during the first cycle among the experiments of Fig. 11 under an air flow rate of (a) 500 sccm; (b) 2500 sccm.

flow rates of 500, 1000 and 2500 sccm (gray, blue and brown
curves in Fig. 11a, respectively). These calculations with respect
to lmoles O2 released/consumed per g of Co3O4 vs. the maximum
one corresponding to stoichiometric reduction/oxidation of 1 g of
Co3O4 (2076 lmoles of O2) are shown in Table 2. Even being
approximate, the values of Table 2 show a definite trend: they
are higher in case of small quantities of powders, and with loaded
porous structures show a tendency for reduction with increasing
flow rate. The highest among the values of coated honeycombs
for an air flow rate of 500 sccm (1086–1354 lmoles of O2/g of
Co3O4) lie in exactly the same range with those reported recently
(Karagiannakis et al., 2016) for a variety of Co3O4-based porous
honeycombs (1090–1210 lmoles of O2/g of Co3O4).
All the values of lmoles of O2 produced/consumed per g of
redox oxide are lower than the theoretically maximum ones. It is
worth mentioning that such lower-than-stoichiometry values have
been reported consistently in the literature in studies other than
TGA ones; first in studies of our group employing solar-irradiated
Co3O4 powders in a rotary kiln (Neises et al., 2012) where they
were attributed to partial reaction of the solid under the particular
experimental conditions. As already mentioned, similar values not
exceeding 60% of theoretical were reported by another research
group for Co3O4- and Mn2O3-made porous pellets of mass in the
range of 8–10 g (Karagiannakis et al., 2014; Pagkoura et al., 2014)
in test rigs similar to the one tested in this study as well as in
the latest work of this group with Co3O4-based extruded and
coated honeycombs. Since in all these studies similar oxygen sen-
sors were employed for the oxygen measurements, there is a pos-
sibility that this consistent lower-than-stoichiometric conversions
could be due to the particular way of measurement which is char-
acterized by a high ‘‘noise” level as well (Fig. 11); however in the
last two references above the sensors’ signal was cross-checked
vs. that of other measurement techniques like oxygen analyzers
and no discrepancy was found. Thus the difference between TGA
 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694 693

Table 2
Calculated amount of oxygen evolved/uptaken per gram of redox powder as percentage of the theoretical from complete stoichiometric reduction/oxidation reactions, occurring
from integration of the respective oxygen concentration curves in experiments with powders and with Co3O4-coated supports.

Powder Mass (g) Air flow rate (sccm) Dwell (h) Oxygen
Reduction Oxidation
lmoles O2/g oxide (%) Of stoichiometric lmoles O2/g oxide (%) Of stoichiometric
Powders, 1 cycle, dwell (Fig. 3)
Co3O4 8 80 2 1477 71 1794 86
Co3O4 10 80 2 1614 77 1757 84
Mn2O3 9 80 2 730 70 X X

Peak no Air flow rate (sccm)

500 1000 2500
lmoles O2 g Co3O4 (%) Of stoichiometric lmoles O2/g Co3O4 (%) Of stoichiometric lmoles O2/g Co3O4 (%) Of stoichiometric
3 Co3O4 – coated honeycombs (m Co3O4 = 75 g), 4 cycles, no dwell (Fig. 11a)
1 (red) 1145 55 827 40 625 30
2 (ox) 1107 53 1112 53 877 42
3 (red) 1210 58 839 40 729 35
4 (ox) 1086 52 1119 54 892 43
5 (red) 1294 62 863 42 536 26
6 (ox) 1110 53 1130 54 833 40
7 (red) 1354 65 880 42 565 27
8 (ox) 1148 55 1130 54 788 38

results that demonstrate stoichiometric O2 production/consump-
tion for a variety of conditions and redox-oxide-coated and –made
porous supports for the two systems has to be attributed else-
where. Future experiments with porous objects of varying porosity
(powders, granules, pellets, honeycombs and foams) are scheduled
to investigate this issue.
4. Conclusions
The target of this work was to advance further the concept of
regenerative ‘‘hybrid” sensible/thermochemical storage of solar
energy proposed for air-operated Solar Thermal Power Plants,
exploring the concept of Cascaded ThermoChemical Storage
(CTCS). In this approach, cascades of porous structures incorporat-
ing different redox oxide materials are rationally distributed along
the storage module to maximize its energy storage density.
In previous works the Co3O4/CoO and Mn2O3/Mn3O4 redox pairs
have been identified as the most promising single-oxide systems.
In parallel, small-scale Co3O4-coated porous honeycombs and
foams have demonstrated stoichiometric, reversible, cyclic redox
operation up to 100 cycles in a thermogravimetric analyzer. Fol-
lowing up these facts, the present work involved the preparation
of larger, lab-scale (£ 25 mm), Co3O4- and Mn2O3-coated porous
cordierite honeycombs and foams and their testing in one- or
two-oxides cascaded configurations. Such structures could be
coated with a high amount of redox oxides, exceeding 100% and
in some cases reaching even 300% of their initial weight.
Parametric studies involved tests with redox oxide quantities in
the range 15–150 g, investigating the effects of air flow rate and
kind and quantity of oxide loaded on the porous supports on the
oxygen release/uptake and the thermal effects of the thermochem-
ical reactions on the air stream exit temperatures. In the case of
Co3O4-coated porous honeycombs and foams, thermochemical
storage was clearly demonstrated as heat uptake/release at con-
stant temperature level, under proper testing conditions (adequate
redox material quantity and high flow rates to overcome heat
losses).
Furthermore, such hybrid sensible/thermochemical storage sys-
tems were compared to respective sensible-only ones. Again, for
the Co3O4-coated honeycombs the additive effect of thermochem-
ical on sensible storage could be qualitatively visualized on the
temperature of the air exiting the thermochemical reaction zone.
The operating conditions of a cascade based on these two oxides
are dictated by the redox behaviour of the Mn2O3/Mn3O4 pair since
this system is characterized by a higher reduction temperature,
lower oxygen uptake/release and reaction heat effects, slower oxi-
dation rate and a large temperature gap between reduction and
oxidation temperature. Under properly selected conditions the
two oxide systems can operate in a complementary mode. During
heating their reduction reactions overlap whereas during cooling
and cycling the oxidation of the reduced state of the oxides takes
place in a consecutive way: that of CoO during cooling and that
of Mn3O4 almost exclusively during re-heating, in different tem-
perature ranges. However, even though this proof-of-concept of
the cascaded configuration was in principle demonstrated, any
beneficial thermal effects of the Mn2O3/Mn3O4 pair system were
not evident, at least in the present configuration tested. The pecu-
liarities of this system need further elaboration of the operation
strategy in combination with larger-scale, properly insulated sys-
tems to allow full exploitation of any thermochemical benefits.
Acknowledgements
Dr. C. Agrafiotis would like to thank the European Commission
for funding of this work within the Project ‘‘FP7-PEOPLE-2011-IEF,
300194; Thermochemical storage of solar heat via advanced reac-
tors/heat exchangers based on ceramic foams (STOLARFOAM)”
under the Intra-European Fellowship Marie Curie Actions of the
7th Framework Programme and to DLR Programmdirektion
Energie (PD-E) for funding through the Project ‘‘Thermochemical
storage for CSP-applications based on Redox-Reactions – from
materials to processes (REDOXSTORE)”.
References
Agrafiotis, C., Becker, A., Roeb, M., Sattler, C., 2015a. Hybrid sensible/
thermochemical storage of solar energy in cascades of redox-oxide-pair-based
porous ceramics. In: ASME 2015 9th International Conference on Energy
Sustainability. American Society of Mechanical Engineers, Paper Number
ES2015-49334.
Agrafiotis, C., de Oliveira, L., Roeb, M., Sattler, C., 2016a. A solar receiver-storage
modular cascade based on porous ceramic structures for hybrid sensible/
thermochemical solar energy storage. AIP Conf. Proc. 1734, 050001.
Agrafiotis, C., Pitz-Paal, R., 2013. Cascaded thermochemical storage of solar heat via
redox oxide materials/Stufenweiser thermochemischer Speicher von
Solarwärme durch Redox-Materialien. German Patent Organization, Patent
disclosure No DE102013211249A1, Germany.
694 C. Agrafiotis et al. / Solar Energy 139 (2016) 676–694
Agrafiotis, C., Roeb, M., Sattler, C., 2016b. Exploitation of thermochemical cycles
based on solid oxide redox systems for thermochemical storage of solar heat.
Part 4: Screening of oxides for use in cascaded thermochemical storage
concepts. Sol. Energy 139, 695–710.
Agrafiotis, C., Roeb, M., Schmücker, M., Sattler, C., 2015b. Exploitation of
thermochemical cycles based on solid oxide redox systems for
thermochemical storage of solar heat. Part 2: Redox oxide-coated porous
ceramic structures as integrated thermochemical reactors/heat exchangers. Sol.
Energy 114, 440–458.
Agrafiotis, C., Tescari, S., Roeb, M., Schmücker, M., Sattler, C., 2015c. Exploitation of
thermochemical cycles based on solid oxide redox systems for thermochemical
storage of solar heat. Part 3: Cobalt oxide monolithic porous structures as
integrated thermochemical reactors/heat exchangers. Sol. Energy 114, 459–475.
Agrafiotis, C., Tsetsekou, A., 2000a. The effect of powder characteristics on washcoat
quality. Part I: Alumina washcoats. J. Eur. Ceram. Soc. 20, 815–824.
Agrafiotis, C., Tsetsekou, A., 2000b. The effect of powder characteristics on washcoat
quality. Part II: Zirconia, titania washcoats—multilayered structures. J. Eur.
Ceram. Soc. 20, 825–834.
Agrafiotis, C., Zygogianni, A., Pagkoura, C., Kostoglou, M., Konstandopoulos, A.G.,
2013. Hydrogen production via solar-aided water splitting thermochemical
cycles with nickel ferrite: experiments and modeling. AIChE J. 59, 1213–
1225.
Alexopoulos, S., Hoffschmidt, B., 2010. Solar tower power plant in Germany and
future perspectives of the development of the technology in Greece and Cyprus.
Renew. Energy 35, 1352–1356.
Carrillo, A.J., Serrano, D.P., Pizarro, P., Coronado, J.M., 2014. Thermochemical heat
storage based on the Mn2O3/Mn3O4 redox couple: influence of the initial
particle size on the morphological evolution and cyclability. J. Mater. Chem. A 2,
19435–19443.
Carrillo, A.J., Serrano, D.P., Pizarro, P., Coronado, J.M., 2015. Improving the
thermochemical energy storage performance of the Mn2O3/Mn3O4 redox
couple by the incorporation of iron. ChemSusChem 8, 1947–1954.
de Gracia, A., Cabeza, L.F., 2015. Phase change materials and thermal energy storage
for buildings. Energy Build. 103, 414–419.
Fricker, H.W., 2004. Regenerative thermal storage in atmospheric air system solar
power plants. Energy 29, 871–881.
Karagiannakis, G., Pagkoura, C., Zygogianni, A., Lorentzou, S., Konstandopoulos, A.,
2014. Monolithic Ceramic redox materials for thermochemical heat storage
applications in CSP plants. Energy Proc. 49, 820–829.
Karagiannakis, G., Pagkoura, C., Halevas, A., Baltzopoulou, P., Konstandopoulos, A.G.,
2016. Cobalt/cobaltous oxide based honeycombs for thermochemical heat
storage in future concentrated solar power installations: multi-cyclic
assessment and semi-quantitative heat effects estimations. Sol. Energy 133,
394–407.
Kosmulski, M., 2009. Compilation of PZC and IEP of sparingly soluble metal oxides
and hydroxides from literature. Adv. Colloid Interf. Sci. 152, 14–25.
Kostoglou, M., Lorentzou, S., Konstandopoulos, A.G., 2014. Improved kinetic model
for water splitting thermochemical cycles using nickel ferrite. Int. J. Hydrogen
Energy 39, 6317–6327.
Li, Y., Guo, B., Huang, G., Kubo, S., Shu, P., 2015. Characterization and thermal
performance of nitrate mixture/SiC ceramic honeycomb composite phase
change materials for thermal energy storage. Appl. Therm. Eng. 81, 193–197.
Neises, M., Tescari, S., Oliveira, L.d., Roeb, M., Sattler, C., Wong, B., 2012. Solar-
heated rotary kiln for thermochemical energy storage. Sol. Energy 86, 3040–
3048.
Pagkoura, C., Karagiannakis, G., Zygogianni, A., Lorentzou, S., Konstandopoulos, A.G.,
2015. Cobalt oxide based honeycombs as reactors/heat exchangers for redox
thermochemical heat storage in future CSP plants. Energy Proc. 69, 978–987.
Pagkoura, C., Karagiannakis, G., Zygogianni, A., Lorentzou, S., Kostoglou, M.,
Konstandopoulos, A.G., Rattenburry, M., Woodhead, J.W., 2014. Cobalt oxide
based structured bodies as redox thermochemical heat storage medium for
future CSP plants. Sol. Energy 108, 146–163.
Ströhle, S., Haselbacher, A., Jovanovic, Z.R., Steinfeld, A., 2016. The effect of the gas-
solid contacting pattern in a high-temperature thermochemical energy storage
on the performance of a concentrated solar power plant. Energy Environ. Sci. 9,
1375–1389.
Tamme, R., Taut, U., Streuber, C., Kalfa, H., 1991. Energy storage development for
solar thermal processes. Sol. Energ. Mater. 24, 386–396.
Tescari, S., Agrafiotis, C., Breuer, S., de Oliveira, L., Neises-von Puttkamer, M., Roeb,
M., Sattler, C., 2014. Thermochemical solar energy storage via redox oxides:
materials and reactor/heat exchanger concepts. Energy Proc. 49, 1034–1043.
Wokon, M., Kohzer, A., Benzarti, A., Bauer, T., Linder, M., Wörner, A., Thess, A., 2014.
Thermochemical energy storage based on the reversible reaction of metal
oxides. In: 3rd International Conference on Chemical Looping, September 9–11,
Göteborg, Sweden.
Zanganeh, G., Khanna, R., Walser, C., Pedretti, A., Haselbacher, A., Steinfeld, A., 2015.
Experimental and numerical investigation of combined sensible–latent heat for
thermal energy storage at 575 °C and above. Sol. Energy 114, 77–90.
Zunft, S., Hänel, M., Krüger, M., Dreißigacker, V., 2014. A design study for
regenerator-type heat storage in solar tower plants – results and conclusions
of the HOTSPOT project. Energy Proc. 49, 1088–1096.
Zunft, S., Hänel, M., Krüger, M., Dreißigacker, V., Göhring, F., Wahl, E., 2011. Jülich
solar power tower—experimental evaluation of the storage subsystem and
performance calculation. J. Sol. Energy Eng. 133, 031019.