Inexpensive thermochemical energy storage utilising additive enhanced limestone

Article abstract of DOI:10.1039/d0ta03080e

Energy storage is one of the key challenges in our society to enable a transition to renewable energy sources. The endothermic decomposition of limestone into lime and CO₂is one of the most cost-effective energy storage systems but it significantly degrades on repeated energy cycling (to below 10% capacity). This study presents the first CaCO₃system operating under physical conditions that mimic a real-life ‘thermal battery’ over an extended cycling life. These important results demonstrate that a thermal energy storage device based on CaCO₃will be suitable for a range of applications,e.g.concentrated solar power plants, wind farms, photovoltaics, and excess grid energy. The operating temperature of 900 °C ensures a higher Carnot efficiency than state-of-the-art technologies at a fraction of the material cost. The capacity degradation of pure CaCO₃as a function of calcination/carbonation cycling is overcome by the addition of either ZrO₂(40 wt%) or Al₂O₃(20 wt%), which results in 500 energy storage cycles at over 80% capacity. The additives result in the formation of ternary compounds,e.g.CaZrO₃and Ca₅Al₆O₁₄, which restrict sintering and allow for the transmission of Ca²⁺and O^2-ions to reaction sites.

Full text of DOI:10.1039/d0ta03080e

Journal of
Materials Chemistry A
View Article Online
View Journal
PAPER
Inexpensive thermochemical energy storage
utilising additive enhanced limestone†
Cite this: DOI: 10.1039/d0ta03080e
*
*
Kasper T. Møller,
Ainee Ibrahim,
Craig E. Buckley
and Mark Paskevicius
Energy storage is one of the key challenges in our society to enable a transition to renewable energy
sources. The endothermic decomposition of limestone into lime and CO2 is one of the most cost-
effective energy storage systems but it significantly degrades on repeated energy cycling (to below 10%
capacity). This study presents the first CaCO3 system operating under physical conditions that mimic
a real-life ‘thermal battery’ over an extended cycling life. These important results demonstrate that
a thermal energy storage device based on CaCO3 will be suitable for a range of applications, e.g.
concentrated solar power plants, wind farms, photovoltaics, and excess grid energy. The operating
temperature of 900 ꢀC ensures a higher Carnot efficiency than state-of-the-art technologies at
a fraction of the material cost. The capacity degradation of pure CaCO3 as a function of calcination/
carbonation cycling is overcome by the addition of either ZrO2 (40 wt%) or Al2O3 (20 wt%), which results
in 500 energy storage cycles at over 80% capacity. The additives result in the formation of ternary
compounds, e.g. CaZrO3 and Ca5Al6O14, which restrict sintering and allow for the transmission of Ca2+
and O2ꢁ ions to reaction sites.
Received 18th March 2020
Accepted 27th April 2020
DOI: 10.1039/d0ta03080e
rsc.li/materials-a
form a shell-like structure around unreacted CaO.5,7,8 As such,
research must be directed towards overcoming the technical
Introduction
challenges associated with using CaCO3 as an energy storage
The relatively low operating temperature and low energy density
(<413 kJ kgꢁ1) of molten salt technology results in a high
concentrated solar power (CSP) energy storage cost. Despite
having energy storage costs lower than Li-ion batteries,
improvements are required to enable higher operating
temperature and efficiency at a lower cost.1 Thermal batteries
offer a direct alternative to electrochemical batteries for excess
renewable energy storage and load levelling.
Various materials have been suggested as the successor to
molten salt,2 including gas–solid based thermochemical energy
storage (TCES) materials such as metal carbonates, which have
high energy densities (>1000 kJ kgꢁ1) making them attrac-
tive.1,3,4 Utilisation of CaCO3 as a TCES material was proposed in
the 1970s.5 However, the major technical issue that has not yet
been adequately solved is the degradation of its CO2 storage
capacity over many cycles of gas release and absorption when
used for heat storage. A previous study highlights how the CO2
capacity in CaCO3 drops to only ꢂ8% of its initial capacity aer
500 cycles, which is exacerbated by high calcination tempera-
tures.6 The cause of the capacity loss has been attributed to:
a loss of porosity in formed CaO, sintering of CaCO3, and
limited CO2 diffusion through CaCO3, which has been shown to
ꢀ
material near 900 C. The 1 bar CO2 equilibrium pressure is at
890 C for CaCO3, dictated by the thermodynamics:9
ꢀ
ꢀ
CaCO3 4 CaO + CO2(g), DH890 ¼ 165.7 kJ;
C
DS890 ¼ 143.0 J Kꢁ1; DG890 z 0 kJ
(1)
ꢀ
ꢀ
C
C
Predominantly, research has focused on using metal oxides
and other minerals for carbon capture rather than thermal
energy storage.4,10 Most studies only investigate the reactions
between CaO and carbon dioxide at low concentration and gas
pressure (#1 bar), usually at moderate temperatures (650–850
ꢀC) for CO2 sequestration purposes.11 Hence, the conditions are
not suitable for a thermal battery as described here, and the
development of TCES systems based on these materials has
been investigated to a lesser degree.12 As such, it is critical to
assess the CaCO3 thermochemical system at operating
temperatures above 890 ꢀC and CO2 pressures above 1 bar since
the rate of CO2 release (and energy release) is strongly depen-
dent on gas pressure,13 while temperature greatly inuences the
system's reversibility.14 The key issue of CaO and/or CaCO3
sintering at high temperatures could be minimised though the
addition of additives to separate active components and mini-
mise agglomeration, for instance through the addition of
MgO.15 Other additives are known to react with either CaCO3 or
CaO at high temperature to form ternary compounds, e.g.
CaSiO3 and CaZrO3.16,17 Cyclic stability has been achieved in the
Fuels and Energy Technology Institute, Curtin University, GPO Box U1987, Perth 6845,
WA, Australia. E-mail: kasper.moller@curtin.edu.au; mark.paskevicius@gmail.com
† Electronic supplementary information (ESI) available. See DOI:
10.1039/d0ta03080e
This journal is © The Royal Society of Chemistry 2020
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
CaCO3 system for sequestration purposes aer the addition of Al2O3 (20 wt%) was cycled for 500 cycles while varying the
aluminium oxide, iron oxide, or zirconium oxide,11,18,19 where calcination/carbonation times between 20 minutes/30
binary compounds appear to act as active catalysts. However, minutes and up to 12 hours each. The carbonation time
the operating conditions oen involve: (i) low-temperature was further extended up to 96 hours to ‘regenerate’ the
carbonation (T < 750 ꢀC) that would restrict agglomeration sample.
and thus do not present the same obstacles as faced in this
Finally, the data were treated manually and compressibility
study and (ii) few cycles (<20), which are not representative of factors were extracted from REFPROP.21
a long-term energy storage solution.
Powder X-ray diffraction
Experimental
Sample preparation
X-ray diffraction (XRD) on powdered samples was performed on
a Bruker D8 Advance diffractometer equipped with a CuKa1,2
source in at-plate geometry mode. Data were collected using
a Lynxeye PSD detector from 15–70ꢀ 2q at 0.02ꢀ steps.
In situ synchrotron radiation powder X-ray diffraction. In
situ time-resolved Synchrotron Radiation X-ray Diffraction
(SR-XRD) data was collected at the Powder Diffraction beam-
line at the Australian Synchrotron, Melbourne, Australia on
CaCO3 was mixed with additives in a 20 wt% ratio (4 g CaCO3 and
1 g of additive), except BaCO3, which was added in a 5 mol%
ratio. 10 mL of ethanol (CH3CH2OH) was added and the mixtures
were ball-milled in stainless steel vials for 2 hours (15 min milling
ꢃ 1 min pause ꢃ 8 reps; 12 ꢃ 8 mm od. stainless steel balls).
Aer ball-milling, the samples were dried in an oven at 105 ꢀC for
approximately 1 hour to obtain a dry powder. Note, that the above
procedure was carried out in an argon-lled glovebox for the Ni
sample, which was dried by applying dynamic vacuum. Addi-
tional ratios were produced for CaCO3–ZrO2 and CaCO3–Al2O3,
i.e. 10 and 40 wt%, hence 4.5 g CaCO3 and 0.5 g additive, or 3 g
CaCO3 and 2 g additive, respectively, were ball-milled using the
same procedure. Finally, a sample of CaCO3–20 wt% ZrO2 (ꢂ4 g
and ꢂ1 g, respectively) was ball-milled for a total of 10 hours
(20 min milling ꢃ 2 min pause ꢃ 30 reps) in a ZrO2 vial using 12
ꢃ 1 mm od. ZrO2 balls, to obtain a small particle size sample.
Various additives were chosen to isolate CaO particles and
prevent sintering (Table S1†).
22,23
˚
a Mythen microstrip detector at l ¼ 0.82502 A.
Powdered
samples were loaded into quartz capillaries (i.d. ¼ 0.5 mm,
o.d. ¼ 0.6 mm), which were attached to a gas system enabling
control of CO2 pressure. The samples were heated by a heat
blower to 950 ꢀC at DT/Dt ¼ 6 ꢀC minꢁ1 while oscillating
during data acquisition. Temperature calibrations were per-
formed using the well-known thermal expansion of NaCl and
Ag.24,25
Scanning electron microscopy
Scanning electron microscopy (SEM) and energy dispersive
spectroscopy (EDS) were performed using a Tescan Mira3
FESEM with an Oxford Instruments X-Max SDD X-ray detector
and AZtec soware. The SEM images were collected using
a backscattered electrons detector, an accelerating voltage of 15
kV, an aperture size of 30 mm, and a working distance of ꢂ15
mm. SEM samples were prepared by embedding powdered
samples in an epoxy resin, which was polished using colloidal
silica. Eventually the polished samples were sputter-coated with
a 10 nm thick carbon layer.
Thermogravimetric and differential scanning calorimetry
Thermogravimetric and simultaneous differential scanning
calorimetry (TG-DSC) were performed on a Mettler Toledo DSC
1 instrument. The samples were heated from room temperature
to 1000 ꢀC (DT/Dt ¼ 10 ꢀC minꢁ1) under an argon ow (20
mL minꢁ1).
Sieverts' method gas studies
Samples were introduced into a SiC sample cell, which was
attached via Swagelok parts to a Hy-Energy PCTpro E&E.20 The
sample was heated to ꢂ900 ꢀC (DT/Dt ¼ 5 ꢀC minꢁ1) at p(CO2) ¼
10ꢁ2 bar, hence decomposing the sample. Subsequently, cycling
of the sample was initiated at isothermal conditions (ꢂ900 ꢀC)
with carbonation at pini(CO2) ꢂ 6 bar for 30 minutes in a 46.3
cm3 volume, followed by calcination at pini(CO2) ꢂ 10ꢁ2 bar for
20 minutes in a 206.7 cm3 volume. A total of 50 cycles was
collected for all samples. Finally, the samples were carbonated
and cooled to room temperature under p(CO2) ꢂ 5 bar. The data
has been corrected to account for the 80 wt% CaCO3 quantity,
which is the active component, while the graphs and fractional
capacity are based on 1 mol of CO2 being released/absorbed
according to reaction scheme (1).
Small angle X-ray scattering
Small angle X-ray scattering data were collected on a Bruker
Nanostar instrument equipped with an Excillium MetalJet
˚
source (GaKa, l ¼ 1.3402 A). Sample powders were pressed
between tape in transmission geometry and measured under
vacuum. Data were background subtracted and put onto an
absolute scale using a NIST SRM3600 glassy carbon standard.26
Specic surface area (SSA) was calculated from the high-q Porod
region (power law slope ¼ 4) using the Unied model in the
Irena soware package for Igor Pro (WaveMetrics).27,28 This is
calculated through:
B
SSA ¼
2pdDr2
Additional experiments were conducted for the additives
ZrO2 and Al2O3, including extended experiments for $100 where B is Porod's constant rened in the unied t, d is the
cycles using the same conditions as above. However, the crystallographic density, and Dr2 is the scattering contrast.
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2020

View Article Online
Paper
Journal of Materials Chemistry A
shows no thermodynamic destabilisation of CaCO3, but some
kinetic modications to the rst calcination, see Fig. S1 and S2.†
A rapid capacity retention screening of the additive-enhanced
CaCO3 (1 hour calcination/carbonation at 900 ꢀC) generally
shows that the CO2 capacity decreases dramatically within the
rst 10 cycles and eventually most of the samples retain a lower
cyclic capacity than the pristine CaCO3 sample due to side reac-
tions. Pristine CaCO3 reaches ꢂ14% of the theoretical CO2
capacity aer 50 calcination/carbonation cycles, similar to
previous studies.29 A few of the additive-enhanced CaCO3 systems
show promise to improve the cyclic stability of the CaCO3.
The SiO2 and NaY additives react with CaCO3 to form spur-
rite (Ca5(SiO4)2CO3), and during cycling they both stabilise at
z16–20% of the expected CO2 capacity, which is slightly better
than the pure CaCO3 sample. Graphite addition aids in a slower
CO2 capacity loss, however, aer 50 cycles the capacity is similar
to SiO2 and NaY at 20%. See ESI† for further information on
these systems. Indeed, the most promising additives are Al2O3
and ZrO2, and thus these will be the focus of this study. The
addition of 20 wt% ZrO2 retains a CO2 capacity of ꢂ80% within
the rst 10 cycles, but a steady degradation of the sample is
observed and at the end of 50 one-hour cycles the CO2 capacity
is reduced to ꢂ55%. Similarly, the addition of 20 wt% Al2O3
results in a steady capacity degradation, and aer 50 CO2 cycles,
it reaches ꢂ49% of the expected capacity. It should be
emphasised that the cyclic capacity is heavily reliant on the
kinetics of CO2 release and absorption, and one-hour cycles do
not always allow for complete reactions to occur.
Results & discussion
Cyclic stability for different additives
Eleven different additive-enhanced CaCO3 samples were
prepared as given in Table S1.† The thermal properties and cyclic
CO2 capacity over 50 one-hour cycles are summarised in
Table S2,† and Fig. 1a. Initial thermal analysis of the samples
Optimising the additives
To further improve the CO2 (and energy) storage capacity,
samples of varying weight ratios of ZrO2 (20 & 40 wt%) and Al2O3
(10, 20 & 40 wt%) were investigated, see Table S1.† The addition
of 40 wt% ZrO2 provides superior CO2 capacity compared to
20 wt% ZrO2 (Fig. S3†), but at the detriment of the remaining
quantity of active component (CaCO3). Extended ball-milling
(ten hours instead of two hours) of the 20 wt% ZrO2 sample
was undertaken to result in smaller particles and better mixing.
However, the long ball-milling results in the same trend in CO2
capacity loss as the two-hour ball-milled sample. Hence, an
initially smaller particle size does not inuence the overall CO2
capacity aer 50 cycles.
The optimum quantity of Al2O3 to add was found to be
20 wt% Al2O3 with a CO2 capacity of ꢂ49% aer 50 one-hour
cycles (Fig. S4†). Here, a 10 wt% Al2O3 loading is insufficient
to retain CO2 capacity and a 40 wt% loading causes CO2 capacity
to drop below 10% within the rst 10 cycles. In a similar manner
to ZrO2, starting from either bulk or nanoparticle (13 nm) Al2O3
results in no measurable change in the CO2 capacity loss during
cycling. Hence, using low-cost bulk Al2O3 is preferable in the
application.
Fig. 1 (a) Carbonation data of additive-enhanced CaCO3 systems over
50 calcination–carbonation cycles (tabs/des ¼ 1 hour) at 900 ꢀC,
highlighting Al2O3 (20 wt%) and ZrO2 (20 wt%) as the most promising
additives. All additives are 20 wt% except BaCO3, which was 5 mol%
(see Table S1†). (b & c) Absorption data of CaCO3–Al2O3 (20 wt% bulk) Extended cycling and the inuence of calcination/carbonation
over 500 calcination–carbonation cycles with varying calcination/
carbonation times. The theoretical maximum refers to the CaCO3
time
The optimised CaCO3–Al2O3 (20 wt%) system was CO2 cycled
500 times at 900 ꢀC for varying calcination/carbonation times,
remaining (59.1%) after assuming a full reaction with the 20 wt% Al2O3
additive into Ca5Al6O14 (reaction scheme (3)).
This journal is © The Royal Society of Chemistry 2020
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
see Fig. 1b and c. The capacity retention aer 500 cycles is pressure between the equilibrium pressure and operating
$80% when carbonation times are extended to $12 hours. It pressure. Despite the kinetic complexity, it is clear that the
is important to note that the theoretical maximum (1 mole of calcination of bulk CaCO3 is slower than for the additive
CO2, 100%) is calculated based on the assumption that reac- enhanced samples, i.e. containing ZrO2 and Al2O3, which may
tion 3 fully occurs. However, it is observed that this reaction is be related to the faster diffusion through the ternary
not completed until the 30th cycle. Hence, the CO2 capacity compounds than through CaCO3.
exceeds the theoretical capacity within the rst 30 cycles as
more CaO/CaCO3 is available for reaction during these cycles
Active components and hypothesised reaction mechanism
until Ca5Al6O14 is completely formed. Additionally, the
A comparison of the crystalline compounds in the additive-
response time for gas absorption/desorption is shown by the
enhanced CaCO3 samples and the resulting carbonated
samples aer 50 calcination/carbonation cycles reveals that
non-reversible side reactions occur between the additives and
the CaCO3 at high temperature, partly explaining the decreasing
high CO2 capacity (ꢂ60%) even for one-hour cycles of calci-
nation and carbonation. This is mirrored in the rapid 30 and
20 minute cycles that maintain a 50% CO2 capacity, while
longer cycling times result in capacities from 80–90%. Hence,
CO2 capacity, see Fig. S7–S17.† Fig. S7† shows powder X-ray
the CaCO3–Al2O3 (20 wt%) system shows remarkable energy
diffraction (XRD) data of the CaCO3–ZrO2 (20 wt%) and
CaCO3–Al2O3 (20 wt%) samples, which highlights that a non-
storage properties: response time and capacity, which makes
this system suitable for a thermal battery. Finally, the heat
reversible reaction between CaCO3/CaO and the additive has
release/uptake should be mentioned. Temperature spikes are
occurred, see reaction scheme (2) and (3):
observed from a thermocouple in the sample reactor when
absorption and desorption is initiated with DT ꢄ 1 ꢀC.
CaO(s) + ZrO2(s) / CaZrO3(s)
5CaO(s) + 3Al2O3(s) / Ca5Al6O14(s)
9CaO(s) + 3Al2O3(s) / Ca9Al6O18(s)
(2)
(3)
(4)
Considering the sample mass (ꢂ250 mg), this temperature
excursion suggests rapid heat (energy) release/uptake (see
Fig. S5†), which is dissipated through the thermal conduc-
tivity of CaO/CaCO3, which has recently been reported.30 The
temperature spikes are more distinct on carbonation, prob-
ably due to active counteraction from the furnace to maintain
The ZrO2 and Al2O3 enhanced CaCO3 systems show cyclic
capacities that are greatly enhanced compared to the other
tested additives (Fig. 1). This makes it clear that a simple
additive that only restricts CaO/CaCO3 sintering is not the
key to capacity retention. The important difference is, in fact,
the properties of the as-formed ternary oxides, CaZrO3 and
ꢀ
the temperature at 900 C on calcination.
Reaction kinetics of CaCO3–Al2O3 (20 wt%) calcination/
carbonation
A comparison of the reaction kinetics at the initial, middle, CaxAlyOz. Parallel research studies into CO2 sequestration and
and nal stage of all samples during the 50 one-hour methane reforming have shown the benets of certain additives
calcination/carbonation cycles is given in Fig. S6.† Two including calcium zirconate and aluminate.31–34 The feature that
different reaction stages are observed during CO2 absorption differentiates these ternary oxides from the other additives is
with the reaction kinetics being much faster in the rst step their ability to conduct ions at high temperatures. Closely
compared to the second. This two-stage reaction limiting related Ca12Al14O33 (Mayenite) is reported to be an oxide ion
process has previously been assigned to the diffusion coeffi- conductor,35 and the layered structure of Ca5Al6O14 is hypoth-
cients of CO2 through the CaO layer (rapid) and the formed esised to facilitate Ca2+ mobility.36 O2ꢁ and Ca2+ migration
CaCO3 layer (slow): DCaO ¼ 0.3 cm2 sꢁ1 and DCaCO ¼ 0.003 cm2 through the additive structure can thus improve reaction
3
sꢁ1.5 Hence, as CaCO3 is formed on the surface of the CaO kinetics and be benecial in retaining the CO2 capacity.17,37 The
particles, reaction kinetics decrease due to the lower diffusion possibility of CaO migration is assigned to the low intrinsic
coefficient through CaCO3. Thus, CO2 absorption having to defect formation energy of 1.61 eV to create a Ca-site Schottky-
propagate from the outside to the inside of a particle through type disorder in CaZrO3.38
a CaCO3 shell is prolonged, whereas CO2 release can initiate
Minor quantities of reaction side products, i.e. CaAl2O4 and
from the outside shell of CaCO3 and hence diffuse through Ca3Al2O6 are observed aer 50 CO2 cycles, see Fig. S7.†
CaO, not CaCO3, which is fast. However, in this study the rate Extending the study to 500 cycles reveals the conversion of side
of calcination is much slower than for carbonation, which is products into Ca5Al6O14 and Ca9Al6O18 (reaction scheme (4))
not clearly described by only a CaCO3 shell that inhibits CO2 whilst a small fraction of Mayenite may also be present
absorption. The gradual CO2 desorption would not be affected (Ca12Al14O33, <3 wt% from Rietveld renement, see Fig. S18†),
by a CaCO3 shell and is instead assigned to the low pressure see Fig. 2. Hence, the formation of Ca5Al6O14 and Ca9Al6O18
differential between the equilibrium pressure (1.2 bar at 900 enables high stability in the cyclic capacity of the system
ꢀC) and the operating pressure during CO2 release (ꢂ0.7 bar). compared to other investigated additives.39 The crystal structure
Faster reaction kinetics are observed during carbonation of Ca9Al6O18 consists of Al6O18 rings with a Ca2+ cation inside.
because it is performed at a signicant overpressure (ꢂ5 bar) However, only 72 of 80 available Ca2+ sites are occupied,40 which
compared to the equilibrium pressure. Thus, reaction is hypothesised to enable Ca2+ mobility through this crystal
kinetics can be improved by increasing the differential structure. The reaction mechanism is depicted in Fig. 3.
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2020

View Article Online
Paper
Journal of Materials Chemistry A
Fig. 2 XRD data (l ¼ 1.54056 A˚) of CaCO3–Al2O3 (20 wt%) absorbed
after 500 calcination/carbonation cycles. The formation of the ternary
compounds Ca5Al6O14 and Ca9Al6O18 is evident.
Fig. 4 (a) In situ SR-XRD data (l ¼ 0.82502 A˚) of CaCO3–Al2O3
(20 wt%, bulk) at 917 ꢀC. Markers: triangle (CaCO3); diamond (CaO);
circle (Al2O3); square (Ca–Al–O compounds). The formation of Ca–
Al–O compounds is highlighted by the dotted box. (b) In situ SR-XRD
data of CaCO3–ZrO2 (40 wt%, bulk) at 917 ꢀC. Markers: triangle
(CaCO3); diamond (CaO); circle (ZrO2); squares (CaZrO3). The pressure
profile is indicated to the right of the respective figure. Carbonation
was performed for ꢂ20 min and calcination for ꢂ30 min in a total of 5
cycles. Intensity is indicated as blue: low and red: high.
Fig. 3 Depiction of the reaction mechanism where Ca2+ and/or O2ꢁ
are able to migrate through the ternary compound and react at the
surface with CO2 to form CaCO3 and generate thermal energy. The
ternary compound also prevents sintering by acting as a barrier
between regions of CaCO3 and CaO.
In situ phase analysis during CO2 cycling
Fig. 4b shows the in situ SR-XRD cycling data of CaCO3–ZrO2
(40 wt%) at 917 ꢀC. The immediate formation of CaZrO3 and
rapid depletion of ZrO2 (within ꢂ1 hour/1 cycle) is evident, and
the amount of CaZrO3 quickly reaches ꢂ65 wt% of the sample
(based on Rietveld renement; theoretically 67.1 wt% at full
reaction). Furthermore, the crystallite size of CaCO3 and CaO
doubles (to >200 and >125 nm, respectively, based on Rietveld
renement, see also Fig. S19 and S20†), over the 5 cycles applied
here, which eventually may result in a capacity decrease due to
large crystallites, see Table S3.† Despite the crystallite growth,
the clear formation and consumption of CaO during CO2
cycling shows the rapid response of the system to calcinate/
carbonate.
In situ synchrotron radiation X-ray diffraction (SR-XRD) data of
CaCO3–Al2O3 (20 wt%) was collected at 917 ꢀC during CO2
absorption (5 bar) and desorption (1 bar) cycling. The initial
decomposition of CaCO3 is evident by the formation of Bragg
reections from CaO, see Fig. 4a. Conversely, when CO2 gas is
applied to the system the CaO Bragg reections decrease rapidly
in intensity due to the reformation of CaCO3. Throughout the 5
desorption/absorption cycles, Bragg reections from Ca–Al–O
containing compounds continue to increase in intensity, but do
not completely react in the 5 cycles applied here. The prolonged
formation of Ca–Al–O compounds agrees with the observations
made in the gas sorption measurements, where it takes ꢂ20
cycles before the consumption of Al2O3 is complete.
This journal is © The Royal Society of Chemistry 2020
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
Morphology
Scanning electron microscopy (SEM) was utilised to analyse the
particle morphology of as-milled and CO2 cycled samples, see
Fig. 5. As-milled CaCO3 consists of nely divided particles in the
size range ꢂ2 to 8 mm. The morphology signicantly changes into
ꢀ
a worm-like, porous structure aer CO2 cycling at 900 C, with
small crystalline particles on the surface of the ‘worms’, see
Fig. S21.† The porosity should enable easy CO2 access to the CaO
particles; however, the worm-like morphology may retard
carbonation through CaCO3/CaO core–shell structure formation.
The as-milled CaCO3–Al2O3 sample consists of small particles
(#100 nm), which is assigned to the hardness of Al2O3 that may
assist in creating smaller particles of CaCO3 during milling. Aer
cycling, the Al2O3 sample has turned into a rock-like morphology,
with a degree of porosity, which allows CO2 migration through
the macro-structure. Furthermore, the as-milled CaCO3–Al2O3
has specic regions that are aluminium (Al2O3) rich whereas the
cycled sample shows that aluminium is well distributed aer
thermal treatment. Here, the aluminium is now combined with
calcium in Ca5Al6O14, which separates regions of CaO/CaCO3 and
is thus ascribed to prevent sintering. In the CaCO3–ZrO2 sample,
the particle sizes range from small CaCO3 particles (ꢂ1 to 5 mm)
to larger ZrO2 particles (ꢂ5 to 10 mm). From EDS mapping, the Zr
seems reasonably well distributed in both the as-milled sample
and aer cycling, which is attributed to the fact that zirconium is
present in each sample either as ZrO2 or CaZrO3. The extensive
sintering observed may explain why the initial smaller particle
size, i.e. extended ball-milling, does not inuence the CO2 cyclic
capacity or reaction kinetics signicantly. Small-angle X-ray
scattering (SAXS) was utilised to determine the specic surface
area of the ball-milled and cycled samples, see Fig. S22 and S23
and Table S4.† Generally, the samples show a low specic surface
area between 1 and 5 m2 gꢁ1, which indicates a low degree of
micro- or meso-porosity and thus supports the proposed mech-
anism of O2ꢁ and Ca2+ migration through the solid material
rather than CO2 diffusion through porous channels. Only the as-
milled samples containing nanoparticles of Al2O3 and ZrO2 have
larger specic surface areas of 29(3) and 52(5) m2 gꢁ1, respec-
tively. In these cases, the specic surface area signicantly
decreases during cycling when particle size increases.
Perspectives and the promise of a thermal battery
A cost comparison of the proposed TCES materials, based on
CaCO3, and the state-of-the-art molten salt technology is provided
in Table 1. A steep $3000 per tonne price for ZrO2 makes the price
per terajoule electrical energy in the CaCO3–ZrO2 (40 wt%) system
expensive, and similar to the state-of-the-art molten salts.
However, Al2O3 is more abundant and far more economical, i.e.
$324 per tonne.41 Hence, the materials cost can be reduced by
ꢂ95% per terajoule electrical energy produced if the molten salt is
replaced with CaCO3–Al2O3 (20 wt%). The operating pressure of <6
bar CO2 reduces the engineering challenges and costs, whilst the
CO2 may be stored in a zeolite or activated carbon by phys-
isorption, which avoids the energy penalty of CO2 compression
during storage.42 Supercritical CO2 may also be utilised as the heat
transfer uid at 900 ꢀC,43 which makes it compatible with the
Fig.
5
Scanning electron microscopy data comparing the
morphology of the as-milled samples (left column), the samples
absorbed after 50 CO2 cycles at 900 ꢀC (right column), and energy
dispersive spectroscopy showing the elemental distribution of
aluminium and zirconium in the respective samples (Al: purple; Zr:
yellow). a and b: CaCO3; c–f: CaCO3–Al2O3 (20 wt%); g–k: CaCO3–
ZrO2 (40 wt%). The samples are embedded in epoxy resin and
polished.
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2020

View Article Online
Paper
Journal of Materials Chemistry A
Rankine–Brayton combined cycle or the Stirling engine for
thermal to electrical energy conversion.44 The latter is highly effi-
cient at 900 ꢀC (practically h ꢂ 49%) and will work well on the kW
scale.45 Overall, the high energy density for CaCO3–Al2O3 and its
small footprint enables its utilisation in Stirling dishes, which are
dispatchable concentrating solar thermal power systems that are
ideal for remote areas, e.g. mine sites. Furthermore, a thermal
battery enables seasonal or load-levelling storage of a variety of
renewable energy from, e.g. wind farms, photovoltaics, and fossil
fuel-based plants. A thermal battery based on CaCO3–Al2O3,
maintains an 80–90% capacity up to 500 cycles, with expectations
for a 30-year lifetime. This is comparable to Li-ion batteries, which
typically reaches a capacity of 80%, dened as the batteries cycle
life, aer 1000 to 4500 cycles, corresponding to a lifespan between
7 and 20 years.3,46,47 Finally, a thermal battery based on CaCO3
holds important intrinsic safety features: (i) the chemical reactions
are limited by equilibrium pressure, which prevents the reactions
from exceeding design conditions (ii) hot, corrosive uids, e.g.
molten salt, is not present (iii) the compounds are not ammable.
Conclusions
The CaCO3–Al2O3 system presented here shows incomparable
cyclic stability for CO2 release and uptake over 500 cycles (>80%)
at realistic operating conditions for utilisation in applications.
The enhanced cyclability is assigned to the formation of ion-
conducting Ca–Al–O compounds, e.g. Ca5Al6O14 and Ca9Al6O18.
Due to a high enthalpy of formation for CaCO3, this system is
ideal for thermochemical energy storage. Both CaCO3 and Al2O3
are cheap and abundant materials worldwide, resulting in an
overall materials cost at only a fraction of state-of-the-art molten
salt technologies (<4%). Furthermore, the system has a rapid
response time with 60% of the full energy capacity stored or
released within 1 hour, while the operating temperature of
900 ꢀC ensures a high Carnot efficiency when converting the heat
into electricity. The ndings described here, enables CaCO3 to be
used as a thermal energy storage material in large-scale appli-
cations at realistic operating conditions. It is envisaged that
thermal batteries could cover the requirements for bulk storage
of renewable energy to cover the intermittent nature of the
renewable energy sources and peak hour demand, or even enable
energy production in remote areas outside the electricity grid.
Conflicts of interest
The contents of this disclosure is subject to patent protection as
Australian Provisional Patent Application No. 2019904801.
Acknowledgements
KTM thanks The Independent Research Fund Denmark for
International Postdoctoral Grant 8028-00009B. MP thanks the
Australian Research Council for ARC Future Fellowship
FT160100303. CEB, MP, and KTM acknowledge the Global
Innovation Linkage project for grant GIL73589. CEB also
acknowledges funding from ARC Linkage grant LP150100730.
Nigel Chen-Tan is acknowledged for help in the laboratory. The
This journal is © The Royal Society of Chemistry 2020
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Paper
Powder Diffraction beamline at the Australian Synchrotron, 24 B. R. S. Hansen, K. T. Møller, M. Paskevicius, A.-C. Dippel,
Melbourne, Australia is acknowledged for the allocation of
beamtime. Finally, SEM, PXD, and SAXS research was under-
taken using the Tescan Mira3 EM (ARC LE130100053), the
P. Walter, C. J. Webb, C. Pistidda, N. Bergemann,
M. Dornheim, T. Klassen, J.-E. Jørgensen and T. R. Jensen,
J. Appl. Crystallogr., 2015, 48, 1234–1241.
Bruker D8 Advance XRD instrumentation (ARC LE0775551), and 25 J. Hu, W. Cai, C. Li, Y. Gan and L. Chen, Appl. Phys. Lett.,
the Bruker NanoStar SAXS instrument (ARC LE140100075) at 2005, 86, 151915.
the John de Laeter Centre, Curtin University. Dr Matthew 26 O. Spalla, S. Lyonnard and F. Testard, J. Appl. Crystallogr.,
Rowles is acknowledged for support with SAXS data collection.
2003, 36, 338–347.
27 G. Beaucage, J. Appl. Crystallogr., 1995, 28, 717–728.
28 J. Ilavsky and P. R. Jemian, J. Appl. Crystallogr., 2009, 42, 347–
353.
References
´
´
´
´
29 P. E. Sanchez Jimenez, A. Perejon, M. Benıtez Guerrero,
´
1 T. D. Humphries, K. T. Møller, W. D. A. Rickard,
M. V. Soanos, S. Liu, C. E. Buckley and M. Paskevicius, J.
Mater. Chem. A, 2019, 7, 1206–1215.
J. M. Valverde, C. Ortiz and L. A. Perez Maqueda, Appl.
Energy, 2019, 235, 543–552.
30 J. E. Bird, T. D. Humphries, M. Paskevicius, L. Poupin and
C. E. Buckley, Phys. Chem. Chem. Phys., 2020, 22, 4617–4625.
2 M. Liu, N. H. Steven Tay, S. Bell, M. Belusko, R. Jacob,
G. Will, W. Saman and F. Bruno, Renewable Sustainable 31 C. Zhao, Z. Zhou, Z. Cheng and X. Fang, Appl. Catal., B, 2016,
Energy Rev., 2016, 53, 1411–1432.
196, 16–26.
3 K. T. Møller, D. Sheppard, D. B. Ravnsbæk, C. E. Buckley, 32 K. S. Sultana, D. T. Tran, J. C. Walmsley, M. Rønning and
E. Akiba, H.-W. Li and T. R. Jensen, Energies, 2017, 10, 1645.
D. Chen, Ind. Eng. Chem. Res., 2015, 54, 8929–8939.
4 S. Kumar and S. K. Saxena, Mater. Renewable Sustainable 33 R. Koirala, K. R. Gunugunuri, S. E. Pratsinis and
Energy, 2014, 3, 30.
P. G. Smirniotis, J. Phys. Chem. C, 2011, 115, 24804–24812.
34 P. Xu, Z. Zhou, C. Zhao and Z. Cheng, Catal. Today, 2016,
259, 347–353.
35 M. Lacerda, J. T. S. Irvine, F. P. Glasser and A. R. West,
Nature, 1988, 332, 525–526.
36 M. Ruszak, S. Witkowski, P. Pietrzyk, A. Kotarba and
Z. Sojka, Funct. Mater. Lett., 2011, 04, 183–186.
37 S. C. Hwang and G. M. Choi, Solid State Ionics, 2008, 179,
1042–1045.
5 R. Barker, J. Appl. Chem. Biotechnol., 1973, 23, 733–742.
6 G. S. Grasa and J. C. Abanades, Ind. Eng. Chem. Res., 2006, 45,
8846–8851.
7 V. Manovic and E. J. Anthony, Energy Fuels, 2010, 24, 5790–
5796.
8 J. C. Abanades and D. Alvarez, Energy Fuels, 2003, 17, 308–315.
9 Outukumpu, HSC Chemistry, Houston, 1 edn, 2006, vol. 6.
10 A. A. Olajire, J. Pet. Sci. Eng., 2013, 109, 364–392.
¨
11 A. M. Kierzkowska, R. Pacciani and C. R. Muller, 38 R. A. Davies, M. S. Islam and J. D. Gale, Solid State Ionics,
ChemSusChem, 2013, 6, 1130–1148. 1999, 126, 323–335.
12 A. J. Carrillo, J. Gonzalez-Aguilar, M. Romero and 39 Y. Hu, W. Liu, H. Chen, Z. Zhou, W. Wang, J. Sun, X. Yang,
J. M. Coronado, Chem. Rev., 2019, 119, 4777–4816.
X. Li and M. Xu, Fuel, 2016, 181, 199–206.
13 E. P. Hyatt, I. B. Cutler and M. E. Wadsworth, J. Am. Ceram. 40 P. Mondal and J. W. Jeffery, Acta Crystallogr., Sect. B: Struct.
Soc., 1958, 41, 70–74.
Crystallogr. Cryst. Chem., 1975, 31, 689–697.
14 M. Zhao, Y. Song, G. Ji and X. Zhao, Energy Fuels, 2018, 32, 41 Alcoa, Quarterly earnings Q3, 2019, available at:https://
´
5443–5452.
investors.alcoa.com/nancial-reports/quarterly-earnings/
2019, accessed 11-02-2020.
42 L. Hauchhum and P. Mahanta, Int. J. Energy Environ. Eng.,
2014, 5, 349–356.
43 Y. Ahn, S. J. Bae, M. Kim, S. K. Cho, S. Baik, J. I. Lee and
J. E. Cha, Nucl. Eng. Technol., 2015, 47, 647–661.
44 V. Zare and M. Hasanzadeh, Energy Convers. Manage., 2016,
128, 227–237.
15 Q. Zhu, S. Zeng and Y. Yu, Environ. Sci. Technol., 2017, 51,
552–559.
16 M. Wang and C.-G. Lee, Energy Convers. Manage., 2009, 50,
636–638.
17 R. Angers, R. Tremblay and A. C. D. Chaklader, J. Am. Ceram.
Soc., 1972, 55, 425.
18 C.-T. Yu and W.-C. Chen, Fuel, 2014, 122, 179–185.
19 C.-T. Yu, S.-Y. Chen, W.-C. Chen and P.-H. Chang, US Pat. 45 A. Bayon, R. Bader, M. Jafarian, L. Fedunik-Hofman, Y. Sun,
´
J. Hinkley, S. Miller and W. Lipinski, Energy, 2018, 149, 473–484.
US20150093317A1, 2015.
20 D. A. Sheppard, M. Paskevicius, P. Javadian, I. J. Davies and 46 J. B. Goodenough and K.-S. Park, J. Am. Chem. Soc., 2013, 135,
C. E. Buckley, J. Alloys Compd., 2019, 787, 1225–1237. 1167–1176.
21 E. W. Lemmon, NIST Standard Reference Database 23: 47 G. L. Soloveichik, Annu. Rev. Chem. Biomol. Eng., 2011, 2,
Reference Fluid Thermodynamic and Transport Properties—
REFPROP, 2013.
22 K. S. Wallwork, B. J. Kennedy and D. Wang, AIP Conf. Proc.,
2007, 879, 879–882.
503–527.
48 C. Y. Zhao, Y. Ji and Z. Xu, Sol. Energy Mater. Sol. Cells, 2015,
140, 281–288.
49 U.S. Geological Survey, Mineral Yearbook, 2015, available at:
https://s3-us-west-2.amazonaws.com/prd-wret/assets/
palladium/production/mineral-pubs/stone-dimension/
myb1-2015-stond.pdf, accessed 11-02-2020.
¨
23 B. Schmitt, C. Bronnimann, E. F. Eikenberry, F. Gozzo,
¨
C. Hormann, R. Horisberger and B. Patterson, Nucl.
Instrum. Methods Phys. Res., Sect. A, 2003, 501, 267–272.
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2020