Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Solar Energy 139 (2016) 695–710

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 4: Screening
of oxides for use in cascaded thermochemical storage concepts
Christos Agrafiotis ⇑, 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: Air-operated Solar Tower Power Plants store the excess solar energy during on-sun operation as sensible
Received 22 February 2016 heat in porous solid materials that function as recuperators during off-sun operation. Their storage
Received in revised form 18 April 2016 capacity can be extended by coating or manufacturing the porous heat exchange modules with oxides
Accepted 20 April 2016
of multivalent metals undergoing reduction/oxidation (redox) reactions accompanied by significant heat
Available online 27 April 2016
effects (e.g. Co3O4/CoO, BaO2/BaO, Mn2O3/Mn3O4, CuO/Cu2O). Furthermore, to maximize the amount of
redox material that can be thermochemically exploited efficiently in a given thermochemical reactor vol-
Keywords:
ume, the idea of employing cascades of porous structures, incorporating different redox oxide materials
Solar energy
Cascaded thermochemical heat storage
and distributed in a certain rational pattern in space tailored to their thermochemical characteristics and
Redox pair oxides the local temperature of the heat transfer medium is set forth.
Thermogravimetric analysis studies with the oxides above were performed to identify the most
suitable ones for further cascaded operation. The Co3O4/CoO redox pair has been already proven capable
of stoichiometric, long-term, cyclic reduction–oxidation under a variety of conditions. The Mn3O4/Mn2O3
redox pair was found herein to be characterized by a large ‘‘temperature gap” between reduction
(940 °C) and oxidation (780–690 °C) temperature, whereas the CuO/Cu2O and BaO2/BaO pairs could
not work reproducibly and quantitatively.
Thermal cycling tests with the Co3O4/CoO and Mn3O4/Mn2O3 powders operating together under the
conditions required for complete oxidation of the less ‘‘robust” Mn3O4/Mn2O3, demonstrated that both
powders can be reduced and oxidized in complementary temperature ranges, extending thus the
temperature operation window of the whole storage cascade.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction regenerative thermal energy storage (TES) systems (Glück et al.,


1991; Tamme et al., 1991), sensible heat is the preferred heat stor-
Generation of Solar Thermal Electricity from Concentrating age type: thermal energy is transferred from a hot working fluid to
Solar Power (CSP) plants offers significant advantages from a sys- a solid in the storage unit during the so-called ‘‘charging” stage and
tem perspective, mostly because of its built-in thermal storage from the solid to the cold working fluid during ‘‘discharging” (Zunft
capabilities (International Energy Agency, 2014). In a series of pre- et al., 2011).
vious publications by the authors (Agrafiotis et al., 2014, 2015b; Regenerative sensible heat storage can be ‘‘hybridized” with
Agrafiotis et al., 2015c) the idea of exploiting redox oxide pairs either Latent Heat Storage (LHS) via Phase Change Materials
coated on or shaped into porous structures like honeycombs and (PCMs) (Glück et al., 1991) or with ThermoChemical Storage
foams, as hybrid sensible-thermochemical solar energy storage (TCS): in addition to sensible-only-effects the heat effects of
means in air-operated Solar Thermal Power Plants (STTPs), was phase changes or chemical reactions can be exploited within the
set forth. The concept is based on the storage principles of such same storage module volume. With respect to TCS, among the pos-
plants like the Solar Tower Jülich (STJ) in Germany (Fig. 1a). In sible reversible reactions (e.g. decomposition of metal hydroxides,
such plants, similarly to existing industrial high-temperature carbonates or oxides) with substantial thermal effects, the
utilization of a pair of redox reactions involving solid oxides of
multivalent metals, is most attractive for large-scale deployment
⇑ Corresponding author.
in STPPs using air as the heat transfer fluid. In this case air can
E-mail address: christos.agrafiotis@dlr.de (C. Agrafiotis).

http://dx.doi.org/10.1016/j.solener.2016.04.034
0038-092X/Ó 2016 Elsevier Ltd. All rights reserved.
696 C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710

HTF flow when charging

HTF flow when discharging


4CuO → 2Cu2O+O2 4CuO 2Cu2O+O2
T = 1120 oC T = 1120 oC
ΔH= 64 kJ/molreact ΔH= 64 kJ/molreact

6Mn2O3 → 4Mn3O4+O2 6Mn2O3 4Mn3O4+O2


T = 1000 oC T = 1000 oC
ΔH= 32 kJ/molreact ΔH= 32 kJ/molreact

2Co3O4 → 6CoO+O2 2Co3O4 6CoO+O2


T = 870 oC T = 870 oC
ΔH= 202 kJ/molreact ΔH= 202 kJ/molreact

2BaO2 → 2BaO+O2 2BaO2 → 2BaO+O2


T = 690 oC T = 690 oC
ΔH= 81 kJ/molreact ΔH= 81 kJ/molreact

(a) (c)

(b)
Fig. 1. (a) The Solar Tower power plant of Jülich, schematic of its operating principle and plant layout (from (Zunft et al., 2011)); (b) schematic of operation principle of
Cascaded Latent Heat Storage (CLHS) with five Phase Change Materials (from (Dinter et al., 1991)); (c) schematic adaption of the idea in the case of hybrid sensible-TCS
storage: an exemplary Cascaded ThermoChemical Storage (CTCS) employing four different redox oxide materials, cascaded according to decreasing reduction temperature
showing the principle of operation during charging and discharging.

be used as both the heat transfer fluid and the reactant (oxygen) rate and the powder characteristics). Oxidation of CoO to Co3O4
and therefore can come to direct contact with the storage material takes place at a slightly lower temperature upon subsequent cool-
(oxide) with the two reduction/oxidation reactions producing sim- ing. This pair undergoes stoichiometric, long-term, cyclic redox
ply oxygen-rich or oxygen-lean air. The potential of such oxides operation under a variety of ramp rates without requiring dwelling
was extensively investigated (Wong, 2011). Equilibrium thermo- at either the reduction or the oxidation temperature. Relevant
dynamic studies with the HSC Chemical Reaction and Equilibrium research started from loose powders (Agrafiotis et al., 2014) and
Software (Roine, 2002) of 74 such single-metal solid oxide – air has reached the level of small-size redox-oxide coated honey-
systems, under nominal air at atmospheric pressure as the carrier combs and foams (Agrafiotis et al., 2015b,c; Tescari et al., 2014)
gas combined with economic potential evaluations, have narrowed as well as larger-scale porous structures either made entirely of
the redox pair systems suited to the operating temperature range Co3O4 or being Co3O4-rich composites (Karagiannakis et al.,
of current as well as future air-operated STPPs, to the ones listed 2014; Pagkoura et al., 2015, 2014). Other studies have dealt with
below in increasing order of equilibrium reaction temperature, the Mn3O4/Mn2O3 (Carrillo et al., 2014a,b; Wokon et al., 2014),
together with their respective storage densities (i.e. heat effects the CuO/Cu2O (Alonso et al., 2015) and the Fe2O3/Fe3O4 (Block
per kg of solid oxide reactant) and the maximum weight change and Schmücker, 2016) pair. In addition to single-metal oxides,
with respect to the solid oxide reactant corresponding to stoi- multi-metal combinations of the cations above have been tested
chiometric reactions: to reduce the operation temperature range and/or cost like e.g. of
Co–Fe (Block et al., 2014), Cu–Co (Block et al., 2013), Cu–Cr, Cu–
2BaO2 þ DH ¢ 2BaO þ O2 Tequilibrium ¼ 690  C Mn (Block and Schmücker, 2016), Mn–Fe (Block and Schmücker,
DH ¼ 474 kJ=kg DWstoich ¼ 9:45% ð1Þ 2016; Carrillo et al., 2015). Mixed compositions corresponding to
spinel Ni, Mg and Cu cobaltates were not proved capable of cyclic
2Co3 O4 þ DH ¢ 6CoO þ O2 Tequilibrium ¼ 870  C redox operation (Agrafiotis et al., 2014). Co- or Ca-containing per-
ovskites (Babiniec et al., 2015, 2016) have also been recently
DH ¼ 844 kJ=kg DWstoich ¼ 6:64% ð2Þ
tested.
However TCS as well as LHS share the common characteristic of
6Mn2 O3 þ DH ¢ 4Mn3 O4 þ O2 Tequilibrium ¼ 1000  C taking place isothermally, i.e. at the equilibrium temperature of
DH ¼ 204 kJ=kg DWstoich ¼ 3:38% ð3Þ the oxide material and at the temperature of fusion/solidification
of the PCM respectively. To exploit the full potential of both meth-
4CuO þ DH ¢ 2Cu2 O þ O2 Tequilibrium ¼ 1030  C ods the entire mass of the storage medium should be above or
below this constant temperature (during on- or off-sun operation,
DH ¼ 811 kJ=kg DWstoich ¼ 10:01% ð4Þ
respectively). In other words, throughout the duration of on-sun
Emphasis on such ThermoChemical Storage (TCS) with redox operation, not only the heat supplied from the heat transfer fluid
oxides has been placed on the Co3O4/CoO pair. Upon heating to the storage medium should be sufficient to melt/reduce the
Co3O4 is reduced to CoO at 885–905 °C (depending on the heating entire mass of the storage material but also to maintain it at a
C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710 697

temperature above its melting/reduction temperature; if at some Table 1


region or after some time the temperature drops below the melt- Properties of powders employed as provided by the suppliers.

ing/reduction temperature of the heat storage medium, solidifica- Powder Supplier Purity Size
tion/oxidation will respectively occur and this storage region will Co3O4 Materion Advanced Chemicals 99.5% 325 mesh
become useless. Therefore a common problem for the two routes CuO No 1 Merck Chemicals P96.0 <160 lm
is that when during on-sun operation the entire volume of the stor- CuO No 2 Beijing Cerametek Materials Co., Ltd. 99.9% 200 mesh
age module is not above the melting/reduction temperature Mn2O3 Materion Advanced Chemicals 99.9% 325 mesh
Ba(NO3)2 Alfa Aesar 99.0% 325 mesh
respectively the processes will consequently not take place/cease La2O3 Alfa Aesar 99.0% 325 mesh
in these regions. SrCO3 Alfa Aesar 99.0% 325 mesh
To tackle this problem, in the case of LHS, the so-called Fe2O3 Alfa Aesar 99.0% 325 mesh
‘‘Cascaded Latent Heat Storage” (CLHS) concept has been proposed,
where salts with different melting temperatures are cascaded as
PCMs from the lower operating temperature upwards to the max- was synthesized via solid-state synthesis. The respective quantities
imum operating temperature. A particular combination of five salts of the powder raw materials La2O3, SrCO3 and Fe2O3 were dry-
in a series as shown in Fig. 1b for parabolic trough power plants mixed and calcined at 1050 °C for 10 h, with a heating rate of
was discussed (Dinter et al., 1991). The salt with the highest melt- 5 °C/min. The characteristics of the powders employed as provided
ing temperature is selected for the hot end of the CLHS and the salt by the suppliers are summarized in Table 1.
with the lowest one for the cold one. The idea is conceptually sim- Thermogravimetric Analysis and Differential Scanning
ple: during charging (on-sun operation) the hot heat transfer fluid Calorimetry (TGA/DSC) experiments were performed with a Net-
enters the cascade at a temperature higher than the melting point zsch STA 449 F3 JupiterÒ instrument. In several experiments the
of the MgCl2/KCl/NaCl mixture providing the necessary heat of process gas effluent was diverted to a mass spectrometer (Pfeiffer,
fusion. Even though the heat transfer fluid will eventually come Omnistar) via which the Oxygen concentration was recorded. The
out of this module at a lower temperature, this temperature is details and the experimental protocol have been described in detail
enough to provide the heat of fusion and melt the second material previously. The basic approach is that during heat-up the reduction
of the cascade and so on. In this way a higher utilization of the pos- reactions (forward schemes of reactions (1)–(4) above) start at a
sible phase changes and a more uniform outlet temperature over specific temperature (hereafter to be denoted as Tred) whereas dur-
time can be achieved. The concept has been verified experimen- ing subsequent cool-down the reverse, oxidation reactions take
tally and numerically by DLR in a cascade of three PCMs, namely place at another temperature (hereafter to be denoted as Tox) that
KNO3, KNO3/KCl and NaNO3, where the positive effects of a CLHS can be different than Tred due to either hysteresis phenomena or
compared to a non-cascaded LHS were affirmed (Michels and peculiarities of the particular oxide systems under study as will
Pitz-Paal, 2007). be discussed below. The extent of reduction/oxidation can be mon-
Similarly a simple inspection of reactions (1)–(4) and the regen- itored by comparing the weight loss/gain vs. the stoichiometric
erative heat storage system of STJ (Fig. 1b) has led to the concept of percentages in schemes (1)–(4). Under such conditions, reduction
Cascaded ThermoChemical Storage (CTCS) (Agrafiotis and Pitz- was practically complete for most systems. However, the oxidation
Paal, 2013). Just like the PCMs, various oxides can be cascaded with extent depends on the particular system and can be controlled via
increasing reduction temperature from the cold to the hot end; an two ways: either by prolonged dwell at a particular temperature
exemplary schematic of the operation concept employing the four where the oxidation rate is significant or by slow heating/cooling
oxides of reactions (1)–(4) is shown in Fig. 1c. The analogy to rates during cycling from the upper to the lower temperature limit,
Fig. 1b is obvious: during on-sun operation air coming out of the so that the materials spend enough time within the temperature
first module of the cascade is cooled, but nevertheless its temper- range of sufficiently fast oxidation kinetics. In all experiments,
ature is above the reduction temperature of the second material in unless stated otherwise, the last cooling step to room temperature
the cascade and therefore sufficient to dissociate it also to its has taken place under Argon gas to ‘‘freeze” the reduced state of
reduced form, and so on. In this respect, just like in the case of the oxide and post-analyse it comparatively to the initial, oxidized
CLHS, the full enthalpy of the heat transfer fluid – despite the fluid state via X-ray Diffraction (XRD) performed with a Siemens (Karl-
becoming progressively cooler – can be stored in individual sruhe, Germany) D-500 Kristalloflex X-ray powder diffractometer
‘‘discretized” parts of the cascade and recovered in a similar (Cu Ka radiation) in the diffraction angle (2 theta) range of 10–80°.
fashion in off-sun operation.
The first three parts of the work have dealt with the Co3O4/CoO
redox pair, which due to its favourable characteristics, is consid- 3. Results – discussion
ered as the starting point for the proposed application. The current
work involves the evaluation of other redox oxide pair powders The single-oxide systems tested are presented below in
among those listed above in a more extensive and systematic sequence of decreasing theoretical reduction temperature (Wong,
way than already initially reported (Agrafiotis et al., 2015a). Issues 2011), starting from the higher-temperature end.
like operational temperature range, peculiarities of redox opera-
tion advantages and limitations in view of their suitability for com- 3.1. The CuO/Cu2O redox pair
bined incorporation in such a cascaded concept, are identified.
The reduction temperature for CuO has been calculated at
1030 °C (Wong, 2011). In Fig. 2a the results of TGA experiments
2. Experimental with two CuO powders from different sources where cycles were
performed between 1090 and 1000 °C with ramp rates of 5 and
Single oxides Co3O4, CuO and Mn2O3 were employed as- 1 °C/min and no dwell at either temperature level are shown. The
received from the respective suppliers. To obtain BaO2, Ba(NO3)2 upper temperature limit was selected just slightly above the reduc-
powder was calcined in alumina crucibles in a furnace under air tion temperature of CuO to avoid possible melting of Cu2O (m.p.
atmosphere at 950 °C for 4 h. In addition to single-oxide redox 1232 °C). In order not to prolong extensively the experiment time,
pairs, a perovskite of targeted stoichiometry La0.3Sr0.7FeO3, shown the initial heat-up to 900 °C was performed with a high ramp rate,
to have high, reversible oxygen uptake/release (Evdou et al., 2010), 30 °C/min. During heat-up both CuO powders exhibit intense
698 C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710

(a) 102 1200


Tred = 1040 oC Tox= 1035 C o
Tmax= 1090 oC

100
1000
Tmin= 1000 oC
98
Weight change (%)

800
96

Temperature (oC)
94 Wt. loss:
10.3 %
10.3 % Wt re-gain: 600
5.0 %

92
Wt re-gain:
2.4 %
400
90

88 CuO No1, 1090-1000oC, 1oC/min


CuO No1, 1090-1000oC, 5oC/min 200
, Weight change (%)
CuO No2, 1090-1000oC, 5oC/min , Temperature
86
0 100 200 300 400 500 600
Time (min)

1200
(b) 102
Tred = 1045 oC CuO No1 vs. No2, 1090-1020oC, 5oC/min
Tox= 1030 oC Tmax= 1090 oC
100
Tmin= 1020 oC 1000
t=2 hrs
98

800
Weight change (%)

96 Temperature (oC)

94 600

92
400

90

200
88 Wt. loss: Wt re-gain:
11.8 % 5.0 %

Wt. loss: Wt re-gain: Wt. loss:


Weight change (%)
Wt re-gain: Wt. loss: Wt re-gain:
10.2 % 9.6 % 9.7 % 9.2 % 9.4 % 8.1 % Temperature
86 0
0 100 200 300 400 500 600 700 800 900
Time (min)
Fig. 2. Comparative weight change curves during TGA of the two CuO powders tested: (a) ramp rates of 1 and 5 °C/min without dwell; (b) ramp rate of 5 °C/min and dwell at
oxidation temperature of 1020 °C.

reduction when a temperature of about 1040 °C is reached (very Powder No1, it is clear that the ramp rate did not have any effect
close to the literature value) accompanied by a weight loss of on the extent of re-oxidation within the temperature limits exam-
10.0–10.3% corresponding to complete reduction stoichiometry ined; the weight increase for both ramp rates was limited to only
(scheme (4) above) and recorded almost instantaneously within a about 2.4 wt%. On the contrary, a difference with respect to the
very narrow temperature range. From the two experiments with powder source was observed: at the higher ramp rate of 5 °C/min,
C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710 699

the two powders exhibited almost identical behaviour during respectively), gradual weight loss is taking place in the next cycles
reduction but Powder No 2 recovered a higher percentage of its and both the upper and the lower bounds of the weight curve shift
weight (5.0 wt%) during subsequent re-oxidation. However, it to lower values. In addition, the oxidation extent is always lower
was clear that under these conditions complete re-oxidation of than that of the previous reduction. These trends are common for
CuO could not be achieved. both powders. It seems that this is the ‘‘best” that can be done with
To possibly improve the re-oxidation extent, the approach of CuO since the dwell temperature is only 10–15 °C lower than that
prolonged dwell at the oxidation temperature was tested with of where oxidation starts. An overall weight loss more than the
both powders. Experiments were performed raising further the (expected) stoichiometric was recorded at the end of the experi-
lower temperature limit to 1020 °C – narrowing very much thus ment. This can only be attributed to any volatilization, especially
the temperature span – and setting the dwell time at this (oxida- since XRD analysis of the product (Fig. 3a) indicated the reduced
tion) temperature at 2 h, raising at the same time the number of phase Cu2O (JCPDS card 05-0667) as the only detectable phase
cycles to 5. Since reduction was fast and quantitative no dwell after reduction. In all cases the product was characterized by a sig-
was set for it. The ramp rate between 1090 and 1020 °C was set nificant volume reduction and was highly sintered in the form of
to 5 °C/min. The results under these conditions are shown in small hard globules (Fig. 3b). From all the results for CuO so far,
Fig. 2b and corroborate the trends observed in Fig. 2a. During the it can be concluded that, even under the best possible and practical
first heat-up both powders undergo a fast, complete reduction. conditions it seems that CuO cannot work reproducibly and quan-
During cool-down from the maximum temperature after the first titatively even for few cycles.
step of complete reduction the materials start oxidizing abruptly
from 1030 °C. Initially a steep weight increase is recorded between 3.2. The Mn2O3/Mn3O4 redox pair
when Tox is crossed until when the lower limit plateau temperature
is reached; oxidation is further accomplished during dwell at The calculated reduction temperature for Mn2O3 has been
1020 °C, however with a slower rate and recovering only a percent- reported as Tred = 1000 °C (Wong, 2011); recent experimental stud-
age of the total weight lost during reduction. A first conclusion is ies have identified it at 950 °C with slight deviations above or
that re-oxidation extent was improved for both powders vs. that below that value depending on the heating rate and the Mn2O3
observed in the previous experiment. Furthermore, during the re- source employed (Carrillo et al., 2014b). Thus the first experiment
oxidation steps powder No 2 systematically recovered a higher with Mn2O3 powder was run between 1000 and 550 °C with ramp
percentage of the weight lost during reduction. However, even rate of 10 °C/min without dwell at either temperature, based on
though during the first re-oxidation step the material recovered the behaviour of Co3O4. The results are shown in Fig. 4a (top). Just
almost quantitatively its initial weight lost (9.6 vs. 10.2 wt% like the other oxide systems, the material exhibits intense

(a) 9000
o CuO reactant
o After reduction stage at TGA
8000
o : CuO
7000 * : Cu2O
Intensity (arbitrary units)

*
6000

5000

4000

3000 o
o oo
2000 o o
*
o * o o
1000 *
*
*
0
10 20 30 40 50 60 70 80
Diffraction angle (2 theta) (o)

(b)

Fig. 3. (a) XRD spectra comparison between the as-received CuO powder No 1 and its reduction product after TGA and last cooling step under Argon; (b) same powder on the
TGA holder before (left) and after (right) the experiment where significant shrinkage and densification can be observed.
700 C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710

1100
(a) Mn2O3
100 600
Tred = 940oC
Tox= 660oC Wt. (%)
Wt loss:
Wt re-gain:
T ( oC) 0
3.46 %
98 0.64 %

96 1000-550oC 10 oC/min
1100
Weight change (%)

Temperature (oC)
100 600
Tred = 925oC
o
Tox= 710 C Wt loss:
3.47 %
0
Wt re-gain:
98 3.13 %

96 1000-600oC 5 oC/min

1100
100 600
Tred = 920oC
Tox= 745 C o
Wt loss:
3.62 %
0
Wt re-gain:
98 3.41 %

96 1000-600oC 2 oC/min

0 200 400 600 800 1000 1200


Time (min)

101 1200
(b) Mn2O3, 10 oC/min, effect of Tdwell oxidation

100 1000

Tred = 940oC

99 800
Weight change (%)

Temperature (oC)
Wt loss:
3.68 %

Wt loss:
3.70 %

98 600
Wt re-gain:
3.51 %

97 400

96 Tdwell oxidation: 900oC 200


750oC
720oC
690oC Wt. (%)
650oC T ( oC)
95 0
0 100 200 300 400 500 600 700 800 900 1000
Time (min)
Fig. 4. Comparative weight change curves during TGA of the Mn2O3 powder tested: (a) effect of ramp rate (10, 5 and 2 °C/min) without dwell during cycling; (b) effect of
oxidation dwell temperature with a ramp rate of 10 °C/min, on oxidation performance.

reduction during heatup when a temperature 940 °C is reached, increases, especially when 660 °C are exceeded. Around 870 °C oxi-
accompanied by a weight loss of 3.38% corresponding to the stoi- dation rate reaches a peak; thereafter is reduced again. This trend
chiometry of complete reduction (scheme (3) above) and recorded is repeated during the second cycle, in which, though a lower level
almost instantaneously within a very narrow temperature range of oxidation is observed compared to that during the first cycle.
even before the sample has reached the upper limit temperature This system is not characterized by a narrow temperature differ-
of 1000 °C. However, during cool-down from the upper tempera- ence between reduction and oxidation like the Co3O4/CoO pair
ture limit, a practically recordable oxidation starts to take place (for which is about 20–30 °C) but rather from an oxidation temper-
only when the temperature reached is as low as 840 °C – signifi- ature much lower than its reduction temperature (940 °C), as it
cantly lower than the Tred of 940 °C. As the temperature is further has also been observed in previous studies (Carrillo et al., 2014b).
decreased the rate of oxidation remains low. After the lower tem- Again one way to increase the re-oxidation extent is to increase
perature limit is reached and heat-up starts, the oxidation rate the dwell time in the temperature range where the oxidation rate
C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710 701

is high. The results by lowering the ramp rate from 10 to 5 and weight change curves have been reported to depend on the source
2 °C/min are shown also in Fig. 4a. The extent of re-oxidation of powder (as also observed in the present study with the two CuO
increases with decreasing ramp rate and weight re-gain almost powders mentioned above) attributing the differences in beha-
reaches the one lost during reduction. In addition, the upper parts viour to particular powder characteristics, like particle size distri-
of the weight curves become ‘‘flatter” and the curves tend to be bution and/or morphology (Carrillo et al., 2014b).
more symmetric with respect to cool-down and heat-up. Neverthe- XRD spectra of representative final products from both sets of
less, oxidation is always slower than reduction and the relevant experiments, with and without dwell, are shown in Fig. 5a (with
curves are ‘‘skewed” to the right. the same line colour as their respective TGA weight curves in
To identify more precisely the temperature where oxidation Fig. 4). The practically stoichiometric extent of reduction is evi-
rate is maximum, experiments were performed with ramp rate of dent: the initial product was single-phase Mn2O3 and the reduction
10 °C/min for long dwell times at different lower level plateau tem- product in all cases single-phase Mn3O4. Reduction can also be
perature values. Five representative such experiments are summa- observed visually, since the reduction product is reddish-like in
rized in Fig. 4b; it can be concluded that the oxidation rate for the colour in contrast to the black colour of the reactant (compared
particular Mn2O3 powder under the specific conditions is signifi- on the TGA holder before and after an experiment in Fig. 5b). The
cant in the range 690–750 °C, maximum around 720 °C (olive extent of re-oxidation seems to affect the crystallinity of the final
curve) where the weight gained during re-oxidation is just margin- product: it is interesting to note the lower crystallinity of the prod-
ally lower than that lost during reduction, and minute outside this ucts from both temperature schedules (with/without dwell) under
temperature range above (e.g. either at 900 or at 650 °C). This the conditions where the weight gain was maximized in contrast
range of oxidation onset temperature determined here (depending to the other two experiments where oxidation took place only
on the cooling rate) is in agreement with results reported in the lit- marginally in both cycles. The two powders which were reduced
erature for this system in several studies either in the form of pow- almost stoichiometrically to Mn3O4 during the first reduction but
der (Carrillo et al., 2014a,b; Wokon et al., 2014) or in the form of exhibited very low re-oxidation percentage in the first oxidation
porous pellets (Karagiannakis et al., 2014). It should be also men- (black and grey curves in Fig 4a, b respectively) and thereafter
tioned that the exact temperature ranges and the shape of the ‘‘oscillated” between much lower levels of weight span are of much

(a)
Mn2O3 as-received, before TGA
after TGA between 1000-550 oC, 10oC/min
after TGA between 1000-600 oC, 2oC/min
5000 after TGA between 1000-720 oC (plateau)
after TGA between 1000-900 oC (plateau)
--- Ref. Mn3O4 --- Ref. Mn2O3
Intensity (arbitrary units)

4000

3000

2000

1000

0
10 20 30 40 50 60 70 80
Diffraction angle (2 theta) (o)

(b)

Fig. 5. (a) XRD spectra comparison between the as-received Mn2O3 powder and its reduction product after characteristic TGA experiments of Fig. 4 with last cooling step
under Argon (colour of XRD spectrum same as weight curve in Fig. 4); (b) Mn2O3 powder on the TGA holder before (left) and after (right) an experiment where the reddish
reduced phase of Mn3O4 is clearly visible. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
702 C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710

(a) 1000
BaO2
100 T = 720 oC
720

Weight change (%) 98 Wt loss:

Temperature (oC)
~ 6.8 %
Weight (%) T( oC)
heating/cooling 5oC/min
Wt loss:
~ 8.0 % heating/cooling 2oC/min
96 heating 5oC/min, cooling 1oC/min
0

94
Wt re-gain: Wt re-gain:
2.0 % 3.5 %

92

0 250 500 750 1000 1250 1500


Time (min)

(b) 1000
BaO2
100 T = 720 oC
720

T = 580 oC
98
Weight change (%)

Temperature (oC)
96 0

94 o
Weight T( C)
2oC/min, T low = 300oC
2oC/min, T low = 450oC
2oC/min, T low = 500oC
92 2oC/min, T low = 550oC
2oC/min, T low = 650oC

0 200 400 600 800


Time (min)
Fig. 6. Cyclic TGA of BaO2 powder: (a) effect of heating and cooling rates on weight change; (b) effect of lower temperature limit on weight change with ramp rate of 2 °C/min
without dwell.

higher crystallinity than their respective counterparts (violet1 and (orange curves in Fig. 6a). After an initial weight loss at low tem-
olive curves in Fig 4a, b respectively) which ‘‘oscillated” within prac- peratures possibly due to absorbed moisture or carbon dioxide
tically the full stoichiometric weight span. not detectable by XRD, upon reaching 720 °C reduction takes
place in accordance to literature and just like all other oxides
3.3. The BaO2/BaO redox pair tested, abrupt weight loss is recorded almost instantaneously
within a very narrow temperature range even before the sample
The powder from the calcination of Ba(NO3)2 at 950 °C was has reached the upper limit temperature of 850 °C. The overall
shown by XRD to be single-phase Barium peroxide BaO2 (see maximum weight loss associated with this reduction though is
below). According to the literature (Bowrey and Jutsen, 1978) 6.8%, significantly less than the 9.45% expected from the stoi-
reduction of BaO2 to BaO takes place at 740–760 °C and is com- chiometry of reaction (1). Further heat-up to 850 °C does not have
pleted below 900 °C, whereas oxidation of BaO starts at about an additional effect on the reduction extent and the weight
600–650 °C and continues even at temperatures lower than remains constant. During subsequent cool-down from 850 °C the
400 °C. Thus a first TGA experiment was run between 850 and material starts gaining weight abruptly when the same tempera-
500 °C with a ramp rate of 5 °C/min throughout and no dwell ture of 720 °C is reached, but oxidation rate at some point during
further cooling is almost negligible and the weight remains practi-
1
For interpretation of colour in Figs. 4, 6 and 9, the reader is referred to the web cally constant; during heat-up from the lower temperature limit
version of this article. the weight increases further but the heating rate is too high and
C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710 703

the material reaches the reduction temperature of 720 °C before 950 °C, shown to be single-phase Barium peroxide BaO2 – JCPDS
oxidation is completed. The behaviour is repeatable from step to card 86-0070). The TGA products are of much lower crystallinity
step. Just like in the case of Mn2O3/Mn3O4, to prolong the material compared to the initial powder before TGA, whereas their phase
dwell in the temperature range favourable for oxidation, a lower composition spans a wide range: from single-phase Barium
ramp rate of 2 °C/min was selected and the lower temperature Carbonate (BaCO3 – JCPDS card 05-0318) to Barium Hydroxide
limit was set to 450 °C (blue curves in Fig. 6a). The behaviour is Hydrate(s) (Ba(OH)⁄2H2O – JCPDS cards 39-1304, 26-0154) and Bar-
qualitatively identical but in this case the weight re-gain during ium Oxide-Peroxide ((BaO)0.25(BaO2)0.75 – JCPDS card 85-1768).
oxidation has been increased from 2.0% to 3.5% of the initial The last, oxide phase form was detected only on samples with
weight, being though still far from that lost during the first reduc- the final step performed under air; on the contrary, the common
tion. Thereafter, the behaviour is repeatable from step to step: the characteristic of all products with the last step under Ar is the
weight lost during each reduction is re-gained during each subse- absence of not only the ‘‘reduced” oxide BaO but of any oxide
quent oxidation – but with slightly decreasing extents of both phase. Since the TGA experiments were run with synthetic air, this
reduction and oxidation. A third experiment with heating rate of indicates that after the TGA experiment, the reduced phase of the
5 °C/min and even lower cooling rate of 1 °C/min (grey curves) oxide BaO might have reacted with moisture and/or CO2 from
from the reduction temperature changed the shape of the weight atmospheric air to form BaCO3 or hydrated Ba(OH)2 until the trans-
curves but did not improve the overall weight gain during fer of the samples to the XRD instrument for analysis or even dur-
oxidation. ing the XRD measurement itself, since no extra precautions for
To investigate whether oxidation takes place at lower tempera- avoiding their contact with air were taken (after TGA, each sample
tures at any further extent as reported in the literature, a series of was weighted in atmospheric environment and stored in a closed
experiments was performed with the heating rate of 2 °C/min vile until XRD analysis). This reactivity of BaO vs. steam and carbon
varying the lower temperature limit of the cycle between 300 dioxide has already been reported (Bowrey and Jutsen, 1978).
and 600 °C. The first cycles of these experiments are shown in The combined results from above, i.e. the non-stoichiometric
Fig. 6b. In some of these experiments the upper temperature limit re-oxidation extent as well as the difficulty in avoiding the Barium
was set to 800 °C; this had no effect on the reduction extent. It is oxide hydration/carbonation under realistic conditions where the
clear that oxidation rate is almost negligible during cooling at tem- air flowing through the redox storage medium cannot be com-
peratures lower than 580 °C where the weight remains practi- pletely purified from moisture or CO2, render this pair not strategi-
cally constant; when the lower temperature limit was set at cally suited for this application.
650 °C (magenta curve) the ‘‘flat” segment of the weight curve is
not observed. However, in all cases the over-all weight gain during 3.4. The LaxSr1xFeO3/LaxSr1xFeO3d redox pair
oxidation is of the order of 50% of the initial weight loss during
reduction. Even though after the first step the extents of reduction XRD analysis of the La2O3/SrCO3/Fe2O3 mixture calcined at
and oxidation coincide, they are still much lower than the theoret- 1050 °C for 10 h (Fig. 8a) has shown La0.3Sr0.7FeO3 as the major
ical, stoichiometric one. Similarly to the Mn2O3/Mn3O4 pair, oxida- phase (JCPDS card 82-1964) as well as minor peaks corresponding
tion is always slower than reduction and the relevant curves are to Sr3Fe2O6 (JCPDS card 45-399) in accordance to previous relevant
also ‘‘skewed” to the right. studies (Nalbandian et al., 2009). The TGA results of this sample
The XRD spectra of the products of several TGA experiments cycled twice between 1200 and 800 °C with heating/cooling rate
where the last cooling step was performed either under Ar or of 5 °C/min are shown in Fig. 8b (the initial heat-up to 600 °C
under air are shown in Fig. 7, together with that of the ‘‘initial” was performed with a higher ramp rate of 30 °C/min). In contrast
material before TGA (product of calcination of Ba(NO3)2 at to the single oxides tested above, the material is not characterized

200 Product of TGA 850-550 oC, heatup 5oC/min, cooldown 1oC/min, last step in air
--- Ref BaO0.25(O2)0.755 (JCPS 85-1768)

0
Product of TGA 850-500 oC, 5oC/min, last step in air
200 --- Ref Ba(OH)2*H2O-JCPS 39-1304; --- Ref Ba(OH)2*H2O-JCPS 26-0154
Intensity (arbitrary units)

--- Ref (BaO)0.25(BaO2)0.755-JCPS 85-1768

0
300 Product of TGA 850-300 oC, 2oC/min, last step in Ar
--- Ref Ba(OH)2*H2O-JCPS 39-1304; --- Ref Ba(OH)2*H2O-JCPS 26-0154

0
Product of TGA 850-450 oC, 2oC/min, last step in Ar
400 --- Ref BaCO3-JCPS 05-0378

0
Ba(NO3)2 calcined at 950 oC-4 hrs, before TGA
1000 --- Ref BaO2

500
0
10 20 30 40 50 60 70
Diffraction angle (2 theta) ( o)

Fig. 7. XRD spectra comparison of various TGA products vs. that of the as-synthesized BaO2 powder from calcination of Ba(NO3)2 at 950 °C for 4 h (TGA ‘‘reactant”).
704 C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710

(a) 500
After calcination at 1050oC, 4 hrs
--- Ref. La0.3Sr0.7FeO3; --- Ref Sr3Fe2O6

400

Intensity (arbitrary units)


300

200

100

0
10 20 30 40 50 60 70 80
Diffraction angle (2 theta) (°)

(b) 101
Tmax = 1200 oC
1200

100 1000
Weight change (%)

Temperature (°C)
Wt loss:
1.35 % Tmin= 800 oC
99 800
Wt re-gain:
0.6 %
600
98
400

97
200
La0.3Sr0.7FeO3 1200-800oC, 5oC/min
Weight change (%)

96
Temperature
0
0 100 200 300 400 500 600
Time (min)

Fig. 8. (a) XRD spectrum of the as-synthesized perovskite powder showing La0.3Sr0.7FeO3 as the major phase and minor peaks of Sr3Fe2O6; (b) weight change curves during
cyclic TGA of La0.3Sr0.7FeO3 powder, tested between 1200 and 800 °C with ramp rate of 5 °C/min without dwell.

by a specific reduction and oxidation temperature value where 3.5. Combination of Co3O4/CoO–Mn2O3/Mn3O4 redox pairs
‘‘abrupt” weight loss or gain takes place; instead weight loss is
observed from the early stages of heating and proceeds in a Based on the findings above, the CuO/Cu2O and the BaO2/BaO
quasi-linear mode until the highest temperature is reached. During pairs were excluded from further studies. The perovskite systems
cool-down to 800 °C this trend is reversed: the material gains although being in principle suitable for a cascaded configuration,
weight in the same fashion. Even though the amount gained during need further research especially with respect to their redox reac-
the first cool-down is less than originally lost during the first heat- tions heat effects. This leaves practically only the two pairs
up from ambient temperature to 1200 °C, the behaviour in the next Co3O4/CoO and Mn2O3/Mn3O4 among the single-oxide redox pairs
steps is totally reversible: when cycled between 1200 and 800 °C for further investigation. It should be noted that the reduction tem-
the material gains and loses exactly the same amount of Oxygen peratures of these systems were identified as 895 and 940 °C
upon cooling/heating respectively. This characteristic is in princi- respectively, i.e. both higher than that of the air provided currently
ple favourable for a cascade configuration since the redox reactions from STJ; however a target in future STPPs is to supply air of higher
can take place throughout the cascade irrespective of its particular temperature. The Mn2O3/Mn3O4 pair has a very peculiar behaviour,
temperature. However the overall weight loss is rather small, characterized by a large temperature gap between reduction
something that implies that also the heat effects of these reactions (940 °C) and oxidation (780–690 °C) temperature. In addition,
will be small as well. A similar quasi-linear weight loss/gain has there exists only a very narrow temperature range (690–
been observed in a series of Co-containing perovskites (Babiniec 750 °C) within which Mn2O3 oxidation proceeds with a significant
et al., 2015), where also the weight change for one cycle under rate, this oxidation is in general slow and needs extended dwell at
air was very small and Argon atmosphere was employed to aug- the optimum temperature (range) for completion.
ment it. The highest reaction enthalpy has been calculated from A comparison of the reaction enthalpies of the two systems was
these experiments as 250 kJ/kg-ABO3 for a La0.3Sr0.7Co0.9Mn0.1O3d performed via DSC experiments. Samples of equal weight of Mn2O3
stoichiometry. and Co3O4 powder were cycled under the same conditions, for 2
C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710 705

cycles between 1000 and 600 °C, without dwell, with low ramp were very small. During the first heat up and the first reduction
rates of 2 and 4 °C/min to ensure almost complete oxidation of point – 928 °C for 4 and 917 °C for 2 °C/min – a small, but neverthe-
Mn3O. The ‘‘corrected” DSC curves (subtracting the curves from less visible peak upwards (endothermic) can be observed in both
those of blank experiments under the same conditions) for the graphs. Thereafter, during all other subsequent reductions/
experiments with the two ramp rates of 2 and 4 °C/min are shown oxidations as recorded from weight change, the DSC signals are
in Fig. 9a and b respectively, together with the relevant TGA curves non-distinguishable. Based on that, the only calculations and
(blue for cobalt oxide). Under such conditions, after applying the comparison of heat effects that could be made were for the first
correction, in one hand the baselines of the plots for the two pow- reduction peak. A comparison for the two ramp rates, with the
ders under the same ramp rate practically coincide in regions respective plots truncated after this point and after subtracting
where no reactions accompanied from significant heat effects take the respective baselines is shown in Fig. 10. Just like in the case of
place; in the other hand the intensities of the endothermic and Co3O4, as the ramp rate increases the DSC peaks become sharper
exothermic peaks of the two systems provide a direct qualitative but the heat effects recorded are lower, as an integration of the area
comparison between the enthalpy effects of reactions of equal under the relevant curves demonstrates. The calculated areas
masses of powders. are provided in Table 2; they were converted to kJ/kg to be
The reaction enthalpies for Co3O4 have been calculated from DSC compared to the theoretical value taken as 202 kJ/kg (Wong,
experiments as a function of ramp rate in Part I of this publication 2011) in the last column of the table, ranging between 42% and
series (Agrafiotis et al., 2014) in the range 480–560 kJ/kg, about 57– 68% of it (85–140 kJ/kg). These values are in the same range with
66% of its theoretical. In the case of Mn2O3 with both rates of 2 and reduction enthalpies reported recently (160–170 kJ/kg) obtained
4 °C/min the DSC signals for exothermic/endothermic reactions from similar experiments with the same system (Carrillo et al.,

(a) 101
Co3O4, Mn2O3, 1000-600oC, 2 oC/min
1

T 1000
, TGA
100
0
99
Weight change (%)

800
98 -1

Temperature (oC)

DSC (mW/mg)
97
600 -2
96

95 -3
400

94
-4
93 200
, DSC corr

92 -5
0 200 400 600 800 1000 1200 1400
Time (min)

(b) 101
Co3O4, Mn2O3, 1000-600oC, 4 oC/min
1
T
, TGA 1000
100
0
99
Weight change (%)

800
98 -1
Temperature (oC)

DSC (mW/mg)

97
600 -2
96

95 -3
400

94
-4
93 200
, DSC corr
92 -5
0 100 200 300 400 500 600 700
Time (min)

Fig. 9. Comparison of TGA and corrected DSC curves between the Mn2O3 and Co3O4 powders between 1000 and 600 °C with heatup/cooldown rate of (a) 2 and (b) 4 °C/min.
706 C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710

101 2.0
Mn2O3, 1000-600oC, 2 and 4 oC/min

1000
100 T=917oC

1.5
99 800
Weight change (%)

Temperature (°C)

DSC (mW/mg)
1.0
98
600
97
0.5

400
96
0.0
95
200
T 2 oC/min TGA, 2oC/min DSC, 2oC/min
T 4 oC/min TGA, 4oC/min DSC, 4oC/min
94 -0.5
0 50 100 150 200 250 300 350 400 450
Time (min)

Fig. 10. Comparison of TGA and ‘‘smoothed” DSC curves for Mn2O3 during the first reduction step between 1000 and 600 °C without dwell and heatup/cooldown rates of 2
and 4 °C/min.

Table 2 Mn3O4 recovering almost fully the total weight lost during
Heats of reduction reactions of Mn2O3 calculated per step from the relevant areas reduction, in a reproducible pattern during six consecutive cycles.
under the curves of the DSC experiments, as a function of ramp rate and comparison The reduction product after TGA is shown in Fig. 11a, where again
to the theoretical value (taken as 202 kJ/kg).
the red colour of the Mn oxide after TGA indicates clearly reduction
Powder Ramp rate Calculated area under DH (kJ/kg) DH (% th.) to Mn3O4. The results are shown together with the respective
(°C/min) the 1st peak of DSC mass-spectrometer signals analysing for compound of atomic mass
curve (mW/mg)
units (a.m.u.) of 32 (O2), (bottom half of Fig. 11b). A characteristic
Mn2O3 4 1.41 84.6 41.9% ‘‘magnification” of the second cycle is shown in Fig. 11c. It is clear
2 2.31 138.6 68.6% that the Co3O4/CoO pair uptakes and releases Oxygen during heat-
up and cool-down respectively. However, distinct oxygen peaks for
the Mn2O3 powder, either alone or in the experiment together with
2014b), where the authors also noticed sharp reduction peaks but the Co3O4 can only be observed during its reduction step. During
much broader oxidation peaks, similarly to the present study. It its oxidation, even if a weight gain is clearly recorded, any oxygen
has recently been shown though, that at least in the case of the signal attributable to Mn2O3 re-oxidation is practically indistin-
Co3O4/CoO pair, such deviations from the theoretical heat of reac- guishable in both experiments. During the heating steps the two
tion can occur due to additional reactions accompanied by heat powders are reduced instantaneously, yet consecutively; their
effects that take place during the actual heating/cooling TGA exper- reduction temperatures are close to each other, but nevertheless
iments like the Co3+ spin state change (Block and Schmücker, 2016) distinct, as it can be clearly observed in Fig. 11c.
but are not taken care of in the theoretical calculations. The two-powder configuration was also tested with a ramp rate
To investigate the combined behaviour of these two redox pairs, of 10 °C/min between 1000 and 720 °C with 3 h dwell time at oxi-
at first experiments with quantities of both powers not touching dation. A relevant experiment with Co3O4 was not performed,
each other (Fig. 11a) placed on the TGA holder were performed since, based on the previous results during such a cycle, complete,
under conditions where the Mn2O3/Mn3O4 pair exhibited the best cyclic reversible reduction–oxidation of the Co3O4/CoO pair is also
oxidation performance, either with slow ramp rate and no dwell, or expected. For comparison purposes though, a mixture of the two
with higher ramp rate and extended dwell at the oxidation powders (Co3O4 + Mn2O3) in the same analogies as in the TGA
temperature. experiment above was prepared and comparatively tested under
In Fig. 11b, the weight curves of experiments corresponding to the same schedule to check for any differences in behaviour, possi-
the first case i.e. with a low ramp rate of 2 °C/min between 1000 ble reactions between Co3O4 and Mn2O3 etc. The TGA results are
and 600 °C and no dwell are shown, in comparison to the ‘‘plain” shown in Fig. 12a compared to the respective experiment with
Co3O4, and Mn2O3 powders. In one hand, the differences between ‘‘plain” Mn2O3 (olive curve as in Fig. 4b) under the same conditions.
the two ‘‘plain” powders are clearly distinguishable: in the case The powder mixture, even though it exhibits an abrupt weight loss
of Co3O4 both reduction and oxidation take place almost instanta- during the first heat-up to 1000 °C, gains weight upon further heat-
neously, to the same stoichiometric extend and the weight curves ing, even during dwell at the high-temperature limit of 1000 °C.
are perfectly symmetric in contrast to the ‘‘skewed” ones of Mn2O3 The sample continues to gain weight with a different rate upon
which, however, under these conditions also proceed to almost cooling from 1000 °C until T = 865 °C is reached. At this point the
stoichiometric oxidation. In the other hand it is obvious that the sample’s weight stabilizes and remains at this value throughout
curve corresponding to the two powders placed simultaneously the subsequent dwell period at the lower temperature of 720 °C.
on the TGA holder is a ‘‘superposition” of the curves of the two In the next cycle the phenomena are similar, but much less intense
individual powders. Reduction takes place almost instantaneously, in scale and they tend to attenuate with increasing number of
whereas oxidation in two consecutive steps: one fast due to the cycles, where only minor changes in weight are further recorded.
oxidation of CoO and one much slower due to the oxidation of The XRD spectra of the Co3O4/Mn2O3 powder mixture before and
C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710 707

(a)

101 1100
(b)
100
600
99

Temperature (°C)
Weight change (%)

98
0
97
96
95
94 1000-600oC; 2 oC/min; no dwell
Co3O4
Mn2O3
93
Weight change (%) Co3O4-Mn2O3 non-mixed
Temperature
92 1100 3.0x10
-8

600
-8
0 2.0x10

amu 32, O 2
-8
1.0x10

32 amu, O2
0.0

0 400 800 1200 1600 2000 2400 2800


Time (min)

(c) 101 1100 1.9x10


-8

o
o
T = 870 C
T = 955 C
100 o
T = 930 C
600
99
Weight change (%)

98
0
97
-8
1.5x10

96
Temperature (°C)

95
amu 32, O

94
2

93
Temperature
Weight change (%) 32 amu, O 2
92 1.1x10
-8

550 600 650 700 750 800 850 900 950


Time (min)
Fig. 11. (a) Co3O4 and Mn2O3 powders placed together and non-mixed on the TGA holder before (left) and after (right) cyclic TGA between 1000 and 600 °C with ramp rate
2 °C/min without dwell during cycling, with the reddish reduced phase of Mn3O4 clearly visible after the experiment; (b) comparative weight change curves during such
experiments with Co3O4, Mn2O3, and non-mixed Co3O4 and Mn2O3 powders in conjunction with gas effluent analysis of emissions of amu 32 (O2) via a mass-spectrometer; (c)
‘‘magnification” of the second cycle.

after TGA are shown in Fig. 12b, from where it can be concluded demonstrated in the TGA graph is not capable of significant oxygen
that the two powders have reacted with each other: the product release/uptake.
after TGA is a low-crystallinity, mixed Co–Mn spinel oxide On the contrary the two non-mixed powders exhibited again
(Co,Mn),(CoMn)2O4 (JCPDS card No 14-808) which, as repeatable and ‘‘additive” redox behaviour. The results of the
708 C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710

1200
(a) 101
T = 950 oC
100
Wt loss:
3.54 %
600
99

Temperature (°C)
98
Weight change (%)

Wt re-gain:
0
3.51 %
97

96

95

94 Co3O4-Mn2O3
10 oC/min; Tdwell oxidation: 720oC
93 Mn2O3
Co3O4-Mn2O3, non-mixed
Weight change (%)
Co3O4-Mn2O3, mixed
Temperature
92
0 200 400 600 800 1000 1200 1400
Time (min)

150
(b) Co3O4-Mn2O3 mixture after TGA between 1000-720 oC
--- Ref (Co,Mn)(Co,Mn)2O4 -18-408

100
Intensity (arbitrary units)

50

0
Co3O4-Mn2O3 mixture as-prepared before TGA
--- Ref Co3O4 --- Ref Mn2O3

200

100

0
10 20 30 40 50 60 70 80
Diffraction angle (2 theta) (°)
Fig. 12. (a) Comparative weight change curves during four TGA cycles between 1000 and 720 °C with ramp rate 10 °C/min and dwell times of 1 h at 1000 °C and 3 h at 720 °C,
of Mn2O3, non-mixed and mixed Co3O4 and Mn2O3 powders; (b) XRD spectrum of the mixed powder before and after TGA, with the latter being a mixed Co–Mn spinel (Co,
Mn), (CoMn)2O4.

non-mixed powders are shown together with the mass- panied by a visible Oxygen uptake. Before the end of this isother-
spectrometer signal for O2, in Fig. 13a; characteristic ‘‘magnifica- mal stage, the sample’s weight has been stabilized to a level just
tions” of two steps, one at the beginning and another towards marginally lower than its initial value before reduction. During
the end of the experiment are shown in Fig. 13b, c. It is again clear the next steps the phenomena are exactly repeated. Only during
in the ‘‘magnifications” of Fig. 13b, c that during all heat-up steps the last the isothermal oxidation stage (Fig. 13c) a ‘‘broad” peak
the two powders are reduced instantaneously and consecutively. evidently corresponding to Oxygen uptake from Mn3O4 can be
During the first cool-down after the dwell at the upper tempera- somehow distinguished. The same trend of ‘‘shift” of the weight
ture limit (Fig. 13b), there is a distinct abrupt weight gain accom- curve to lower absolute values with increasing number of cycles
panied by sharp oxygen uptake corresponding to Co3O4 (its re- typical for all experiments involving Mn2O3 can also be observed
oxidation temperature is slightly lower than its initial reduction here. Entirely similar in shape oxygen concentration profiles
temperature); this weight gain is not equal to the total weight loss including very sharp peaks during reduction and very broad and
that occurred during reduction. During the next, isothermal step at low peaks during oxidation have been also obtained in other recent
720 °C, even though a further increase of the sample’s weight is studies with the Mn2O3/Mn3O4 (Karagiannakis et al., 2014)
recorded – obviously due to Mn3O4 oxidation – this is not accom- employing 10 g pellets and dwell at oxidation temperature levels.
C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710 709

(a) 101 1200 4.0x10 -9

100
3.0x10 -9
600

Weight change (%)


99
2.0x10 -9
98

amu 32, O 2
0
97
1.0x10 -9
96
0.0

Temperature ( oC)
95 Wt loss: Wt re-gain:
5.08 % 4.80 %
Wt loss: Wt re-gain:

94 5.04 % 4.73 % Wt loss:


4.54 %
Wt re-gain:
4.69 %
-1.0x10-9
Wt loss: Wt re-gain: Wt loss:
4.99 % 4.74 % 4.88 %

o o
Co3O4-Mn2O3; 10 C/min; Tdwell oxidation: 720 C
93 Mn2O3
Weight change (%)
Temperature
Co3O4-Mn2O3, non-mixed
32 amu, O2
-2.0x10-9
92
0 200 400 600 800 1000 1200 1400
Time (min)
-9 -9

(b) 101
1200 4.0x10 (c) 101 1200 4.0x10

100 100 T = 965 oC


T = 955 oC
99 T = 935 oC o
99
T = 840 C T = 940 oC
Temperature (°C)

Weight change (%) T = 840 oC


Weight change (%)

Temperature (°C)
-9 -9
98 800 2.0x10 98 O2 from Co3O4 800 2.0x10
O2 from Mn2O3

amu 32, O2
97 O2 from Co3O4 97

96 96 O2 from Mn2O3
O2 to Mn3O4

95 400 0.0 95 400 0.0


amu 32, O2

94 94 O2 to CoO
O2 to CoO

Co3O4-Mn2O3; 10 oC/min; Tdwell oxidation: 720oC


93 Mn2O3
Weight change (%) 93 Co3O4-Mn2O3; 10 oC/min; Tdwell oxidation: 720oC Weight change (%)
Temperature Mn2O3 Temperature
Co3O4-Mn2O3, non-mixed Co3O4-Mn2O3, non-mixed
32 amu, O2 -9 32 amu, O2 -9
92 0 -2.0x10 92 0 -2.0x10
60 80 100 120 140 160 180 1000 1100 1200 1300
Time (min) Time (min)

Fig. 13. (a) Same weight change curves as in Fig. 12a in conjunction with gas effluent analysis of emissions of amu 32 (O2) via a mass-spectrometer; (b), (c) ‘‘magnifications”
of the first and last cycles respectively.

4. Conclusions operation of current and future air-operated STPPs, besides


Co3O4/CoO only Mn2O3/Mn3O4 is capable of cyclic operation. How-
The current solar heat storage concept in air-operated Solar ever, whilst the Co3O4/CoO system undergoes stoichiometric, cyclic
Tower Power Plants as sensible heat in porous solid materials oper- reduction–oxidation under a variety of heating/cooling rates and
ating off-sun as regenerators can be rendered to a ‘‘hybrid” sensi- within a narrow temperature range, Mn2O3/Mn3O4 is characterized
ble/thermochemical one via coating the porous heat exchange by a lower reaction heat effect especially during oxidation, slower
modules with oxides of multivalent metals for which their reduc- oxidation rate and a large temperature gap between reduction
tion/oxidation reactions can take place within the plant’s air tem- (940 °C) and oxidation (780–690 °C) temperature. Furthermore,
perature operating range and are accompanied by significant heat the CuO/Cu2O and BaO2/BaO pairs cannot work reproducibly and
effects or by manufacturing them entirely of such oxides. Further- quantitatively due to different reasons. The former is characterized
more, the construction modularity of current state-of-the-art sen- by a reduction temperature high enough to cause extensive sinter-
sible storage systems provides for the implementation of concepts ing and shrinkage of the material. The reduced phase of the latter,
like spatial variation of redox oxide chemistry and solid materials BaO, reacts even with traces of any moisture and/or CO2 present in
porosity along the storage module, to enhance the utilization of the surrounding atmosphere to hydrated Barium hydroxides and
the heat transfer fluid and the storage of its enthalpy. In this per- Barium carbonate respectively, a fact severely limiting the poten-
spective the idea of employing cascades of various porous struc- tial implementation of this pair in the particular application under
tures, incorporating different redox oxide materials and realistic on-site conditions. In the other hand perovskite systems
distributed in a certain rational pattern in space tailored to their that are not characterized by a distinct reduction/oxidation tem-
thermochemical characteristics and to the local air temperature perature range but they release/gain oxygen in a continuous mode
has been set forth and tested. during the heating/cooling stages respectively are in principle
TGA studies were performed among those redox systems iden- favourable for a cascaded configuration like the one mentioned
tified as in-principle suitable via thermodynamic calculations, herein, together with any other current or future mixed oxide
namely CuO/Cu2O, Mn2O3/Mn3O4 and BaO2/BaO as well as with compositions, provided that they can operate in a reversible,
one perovskite system, to determine their operational temperature multi-cyclic redox mode with significant reaction heat effects.
range, cyclability, limitations and peculiarities of redox operation. A combination of two single-oxide systems, namely Co3O4/CoO
These experiments have shown that among the single-oxide and Mn2O3/Mn3O4 has been further explored. Thermal cycling tests
systems exhibiting redox behaviour in the temperature range of with these two powders together under the conditions required for
710 C. Agrafiotis et al. / Solar Energy 139 (2016) 695–710

complete oxidation of the less ‘‘robust” one, Mn3O4/Mn2O3, Babiniec, S.M., Coker, E.N., Miller, J.E., Ambrosini, A., 2016. Doped calcium
manganites for advanced high-temperature thermochemical energy storage.
demonstrated in principle the proof-of-concept of the cascaded
Int. J. Energy Res. 40 (2), 280–284.
configuration, i.e. that both powders can be reduced and oxidized Block, T., Bauer, T., Schmücker, M., 2013. Thermochemical Storage/
in complementary temperature ranges, extending thus the temper- Thermochemischer Speicher, German Patent Organization, Patent disclosure
ature operation window of the whole storage cascade. However No DE102014200803A1, Germany.
Block, T., Knoblauch, N., Schmücker, M., 2014. The cobalt-oxide/iron-oxide binary
the heat effect of Co3O4 per unit volume is much higher than that system for use as high temperature thermochemical energy storage material.
of Mn2O3. Therefore, to maximize the amount of heat generated Thermochim. Acta 577, 25–32.
per unit volume of a cascaded storage module during discharging, Block, T., Schmücker, M., 2016. Metal oxides for thermochemical energy storage: a
comparison of several metal oxide systems. Sol. Energy 126, 195–207.
the region of the cascade containing Mn2O3 should be limited only Bowrey, R.G., Jutsen, J., 1978. Energy storage using the reversible oxidation of
to where the temperature during discharging cannot reach the oxi- barium oxide. Sol. Energy 21, 523–525.
dation temperature of Co3O4; this naturally depends on the flow Carrillo, A.J., Moya, J., Bayón, A., Jana, P., de la Peña O’Shea, V.A., Romero, M.,
Gonzalez-Aguilar, J., Serrano, D.P., Pizarro, P., Coronado, J.M., 2014a.
rate and temperature of the air stream introduced in the cascade Thermochemical energy storage at high temperature via redox cycles of Mn
during the discharging step. The demonstrated chemical reaction and Co oxides: pure oxides versus mixed ones. Sol. Energy Mater. Sol. Cells 123,
between the Co3O4 and Mn2O3 powders to a redox-inert phase 47–57.
Carrillo, A.J., Serrano, D.P., Pizarro, P., Coronado, J.M., 2014b. Thermochemical heat
induces the necessity of some physical separation. This, for exam- storage based on the Mn2O3/Mn3O4 redox couple: influence of the initial
ple, could be just a ‘‘thin slice” of non-coated porous supports particle size on the morphological evolution and cyclability. J. Mater. Chem. A 2,
between the Co3O4- and the Mn2O3-coated ones, operating as 19435–19443.
Carrillo, A.J., Serrano, D.P., Pizarro, P., Coronado, J.M., 2015. Improving the
sensible-only heat storage modules. Such aspects, including the
thermochemical energy storage performance of the Mn2O3/Mn3O4 redox
relevant materials cost, are currently under consideration for the couple by the incorporation of iron. ChemSusChem 8, 1947–1954.
design of a larger-scale unit. Dinter, F., Geyer, M., Tamme, R., 1991. Thermal Energy Storage for Commercial
Application (TESCA). A Feasibility Study on Economic Storage Systems.
Springer-Verlag, Berlin.
Acknowledgements Evdou, A., Zaspalis, V., Nalbandian, L., 2010. La1xSrxFeO3d perovskites as redox
materials for application in a membrane reactor for simultaneous production of
Dr. C. Agrafiotis would like to thank the European Commission pure hydrogen and synthesis gas. Fuel 89, 1265–1273.
Glück, A., Tamme, R., Kalfa, H., Streuber, C., 1991. Investigation of high temperature
for funding of this work within the Project ‘‘FP7-PEOPLE-2011-IEF, storage materials in a technical scale test facility. Sol. Energ. Mater. 24, 240–
300194; Thermochemical storage of solar heat via advanced reac- 248.
tors/heat exchangers based on ceramic foams (STOLARFOAM)” International Energy Agency OECD/IEA, 2014. Technology Roadmap: Solar Thermal
Electricity.
under the Intra-European Fellowship Marie Curie Actions of the Karagiannakis, G., Pagkoura, C., Zygogianni, A., Lorentzou, S., Konstandopoulos, A.,
7th Framework Programme and to DLR Programmdirektion Ener- 2014. Monolithic ceramic redox materials for thermochemical heat storage
gie (PD-E) for funding through the Project ‘‘Thermochemical stor- applications in CSP plants. Energy Procedia 49, 820–829.
Michels, H., Pitz-Paal, R., 2007. Cascaded latent heat storage for parabolic trough
age for CSP-applications based on Redox-Reactions – from solar power plants. Sol. Energy 81, 829–837.
materials to processes (REDOXSTORE)”. Nalbandian, L., Evdou, A., Zaspalis, V., 2009. La1xSrxMO3 (M = Mn, Fe) perovskites as
materials for thermochemical hydrogen production in conventional and
membrane reactors. Int. J. Hydrogen Energ. 34, 7162–7172.
References
Pagkoura, C., Karagiannakis, G., Zygogianni, A., Lorentzou, S., Konstandopoulos, A.G.,
2015. Cobalt oxide based honeycombs as reactors/heat exchangers for redox
Agrafiotis, C., Becker, A., Roeb, M., Sattler, C., 2015a. Hybrid sensible/ thermochemical heat storage in future CSP plants. Energy Procedia 69, 978–
thermochemical storage of solar energy in cascades of redox-oxide-pair-based 987.
porous ceramics. In: ASME 2015 9th International Conference on Energy Pagkoura, C., Karagiannakis, G., Zygogianni, A., Lorentzou, S., Kostoglou, M.,
Sustainability. American Society of Mechanical Engineers, Paper Number Konstandopoulos, A.G., Rattenburry, M., Woodhead, J.W., 2014. Cobalt oxide
ES2015-49334. based structured bodies as redox thermochemical heat storage medium for
Agrafiotis, C., Pitz-Paal, R., 2013. Cascaded Thermochemical Storage of Solar Heat future CSP plants. Sol. Energy 108, 146–163.
via Redox Oxide Materials/Stufenweiser thermochemischer Speicher von Roine, A., 2002. Outokumpu HSC chemistry for windows: chemical reaction and
Solarwärme durch Redox-Materialien, German Patent Organization, Patent equilibrium software with extensive thermochemical database. User’s guide,
disclosure No DE102013211249A1, Germany. version 5.
Agrafiotis, C., Roeb, M., Schmücker, M., Sattler, C., 2014. Exploitation of Tamme, R., Taut, U., Streuber, C., Kalfa, H., 1991. Energy storage development for
thermochemical cycles based on solid oxide redox systems for solar thermal processes. Sol. Energ. Mater. 24, 386–396.
thermochemical storage of solar heat. Part 1: testing of cobalt oxide-based Tescari, S., Agrafiotis, C., Breuer, S., de Oliveira, L., Neises-von Puttkamer, M., Roeb,
powders. Sol. Energy 102, 189–211. M., Sattler, C., 2014. Thermochemical solar energy storage via redox oxides:
Agrafiotis, C., Roeb, M., Schmücker, M., Sattler, C., 2015b. Exploitation of materials and reactor/heat exchanger concepts. Energy Procedia 49, 1034–
thermochemical cycles based on solid oxide redox systems for 1043.
thermochemical storage of solar heat. Part 2: redox oxide-coated porous Wokon, M., Kohzer, A., Benzarti, A., Bauer, T., Linder, M., Wörner, A., Thess, A., 2014.
ceramic structures as integrated thermochemical reactors/heat exchangers. Sol. Thermochemical Energy Storage based on the Reversible Reaction of Metal
Energy 114, 440–458. Oxides, 3rd International Conference on Chemical Looping, September 9–11,
Agrafiotis, C., Tescari, S., Roeb, M., Schmücker, M., Sattler, C., 2015c. Exploitation of Göteborg, Sweden.
thermochemical cycles based on solid oxide redox systems for thermochemical Wong, B., 2011. Thermochemical Heat Storage for Concentrated Solar Power, Final
storage of solar heat. Part 3: cobalt oxide monolithic porous structures as Report for the U.S. Department of Energy, San Diego, CA, U.S.A. <http://www.
integrated thermochemical reactors/heat exchangers. Sol. Energy 114, 459–475. osti.gov/scitech/biblio/1039304/> (last accessed on 16/11/2015).
Alonso, E., Pérez-Rábago, C., Licurgo, J., Fuentealba, E., Estrada, C.A., 2015. First Zunft, S., Hänel, M., Krüger, M., Dreißigacker, V., Göhring, F., Wahl, E., 2011. Jülich
experimental studies of solar redox reactions of copper oxides for solar power tower—experimental evaluation of the storage subsystem and
thermochemical energy storage. Sol. Energy 115, 297–305. performance calculation. J. Sol. Energy Eng. 133, 031019.
Babiniec, S.M., Coker, E.N., Miller, J.E., Ambrosini, A., 2015. Investigation of
LaxSr1xCoyM1yO3d (M = Mn, Fe) perovskite materials as thermochemical
energy storage media. Sol. Energy 118, 451–459.

You might also like