Microstructural control by freeze-casting of CaO architectures for improved and stable thermochemical energy storage performance

Research papers Microstructural control by freeze-casting of CaO architectures for improved and stable thermochemical energy storage performance Nabil Amghar a , Juan Ivorra-Martinez a,b , Antonio Perej´ on a,c,* , Dorian Hanaor d , Aleksander Gurlo d , Joaquín Ramírez-Rico a,e , Luis A. P´ erez-Maqueda a,* , Pedro E. S´ anchez-Jimenez a,* a Instituto de Ciencia de Materiales de Sevilla (C.S.I.C.-Universidad de Sevilla), C. Am´ erico Vespucio 49, Sevilla 41092, Spain b Institute of Materials Technology (ITM), Universitat Polit` ecnica de Val` encia (UPV), Plaza Ferr´ andiz y Carbonell 1, 03801 Alcoy, Alicante, Spain c Departamento de Química Inorg´ anica, Facultad de Química, Universidad de Sevilla, 41012 Sevilla, Spain d Technische Universit¨ at Berlin, Faculty III - Process Sciences, Institute of Material Science and Technology, Chair of Advanced Ceramic Materials, Straße des 17. Juni 135, 10623 Berlin, Germany e Departamento Física de la Materia Condensada, Facultad de Física, Universidad de Sevilla, Avda. Reina Mercedes SN, 41012 Sevilla, Spain ARTICLE INFO Keywords: Freeze-casting Calcium Looping Thermochemical energy storage Porous structures CaCO 3 ABSTRACT This study investigates the development of porous calcium-based monoliths via freeze-casting (FC) as a novel approach for thermochemical energy storage, particularly within the Calcium Looping (CaL) process. The freezecasting technique enabled the fabrication of scaffolds with controlled porosity using polyvinyl alcohol (PVA) as a binder. Experimental results demonstrated that freeze-cast monoliths exhibited superior multicycle performance under various carbonation and calcination conditions. The FC-CaCO 3 monolith achieved the highest residual conversion of 68.1 % under mild vacuum calcination conditions (780 ◦C, 0.1 bar CO 2 ), significantly surpassing other configurations. Tests conducted in an inert atmosphere also yielded favorable results, with a conversion of 56.1 %, outperforming equivalent raw powder samples. The enhanced performance is attributed to improved CO 2 interaction with the porous structure, mitigating sintering effects and preserving active surface area. Morphological observations by X-ray tomography and SEM confirmed limited particle sintering after multiple cycles, maintaining a reactive surface that supported consistent conversion rates. The pore size distribution of the material evolves upon cycling resulting in an increased microporosity, while the pore network maintains a low tortuosity ( τ ~ 1.5–2.0). The addition of dopants such as ZrO 2 and SiO 2 did not enhance performance, as the monoliths' inherent structure provided sufficient stability. These findings highlight freeze-casting as a promising method for creating advanced porous materials suitable for energy storage applications. 1. Introduction The Calcium Looping (CaL) process has been extensively investigated as a promising high-temperature thermochemical energy storage (TCES) method, compatible with concentrating solar power (CSP) plants [1–3]. This process relies on the reversible reaction between CO 2 and CaO to form CaCO 3 , as represented by Eq. (1): CaCO3⇄CaO +CO2ΔHr0= ∓178 kJ/mol (1) The CaL process offers significant advantages, including high energy density and the abundant availability, low cost, and non-toxic nature of the raw materials involved, such as calcium-containing minerals, industrial byproducts, and even waste materials [4–6]. In a typical CSP-CaL operation, solar radiation drives the endothermic decomposition of CaCO 3 into CaO and CO 2 [3,7,8]. These reaction products are then transported to separate storage reservoirs, where they remain until needed. When energy is required, they are recombined in a carbonator reactor, where the reverse exothermic reaction releases heat, which is subsequently harnessed in a power cycle, typically a CO 2 -closed Brayton cycle, to generate electricity [8–10]. For effective calcination, temperatures of approximately 750 to 950 ◦C, contingent on the calcination atmosphere, are essential [11]. While the carbonation reaction proceeds rapidly, its extent is influenced by the CO 2 partial pressure and temperature [12,13]. Carbonation is generally * Corresponding authors. E-mail addresses: [email protected] (A. Perej´ on), [email protected] (L.A. P´ erez-Maqueda), [email protected] (P.E. S´ anchez-Jimenez). Contents lists available at ScienceDirect Journal of Energy Storage journal homepage: www.elsevier.com/locate/est https://doi.org/10.1016/j.est.2025.116681 Received 22 December 2024; Received in revised form 27 March 2025; Accepted 13 April 2025 Journal of Energy Storage 125 (2025) 116681 Available online 8 May 2025 2352-152X/© 2025 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/bync-nd/4.0/ ). conducted at about 800–850 ◦C to achieve maximum conversion rates and high thermoelectric efficiency [14,15]. From a materials perspective, a critical barrier to the large-scale deployment of this technology is the gradual deactivation of CaO, primarily due to the loss of active surface area caused by sintering at the elevated temperatures required by the process [16,17]. Sintering is further accelerated in CO 2 -rich environments [18,19]. Multiple strategies have been employed to mitigate sintering. One such approach involves optimizing reaction conditions to lower calcination temperatures by utilizing an inert atmosphere, reduced pressures, or steam [20–22]. Another method focuses on modifying the storage material by adding stabilizing inert agents such as metal oxides. These additives act as structure-stabilizing agents that mitigate the loss of reactivity due to sintering [23–25]. Sorbent engineering strategies aim to enhance the material's textural properties and create macroporous structures that facilitate carbonation [25–28]. Techniques such as acetic acid modification or the use of sacrificial templates can yield highly porous structures that improve the behavior of limestone [21,29,30]. Finally, small loads of alkali chlorides and carbonate salts have been explored as active additives to enhance cyclic energy storage densities [31–33]. Freeze casting, also known as ice-templating, is a versatile technique that has been widely adopted in recent years to fabricate porous materials from a variety of substances, including ceramics, metals, polymers, and biomacromolecules [34,35]. The process yields structures with distinct textural and morphological properties. During freezing, the solvent, typically water or camphene, solidifies to form a crystalline structure that templates the pores. Upon sublimation of the frozen solvent, channels are left throughout the scaffold, resulting in a porous architecture. Freeze casting can mimic natural cellular structures, imparting specific porosity to materials. Consequently, it is a strategic approach to designing synthetic materials inspired by nature. By controlling the freezing parameters, it is possible to manipulate the internal structure to achieve specific pore morphologies, including cellular, dendritic, and lamellar forms [35,36]. Previous studies have reported that directional freezing techniques can significantly improve the utilization of renewable energy for electrochemical energy storage and conversion, further enhancing the functionality of materials with multiscale porous structures [37]. In directional freeze casting, polyvinyl alcohol (PVA) is commonly used to enhance the dispersibility of ceramic slurries and improve the mechanical strength of the resulting green ceramic bodies. PVA prevents particle agglomeration that can hinder crystal growth and interacts with particles during freezing to suppress excessive crystal formation. Adding PVA improves pore connectivity and open porosity. As a soluble polymer, PVA enhances the mechanical integrity of the ceramic body [38,39]. This study explores for the first time date a freeze casting approach for the synthesis of porous CaO sorbents. Using the innovative freezecasting technique, highly porous CaO architectures were developed. Multicycle performance were evaluated in a laboratory reactor considering various operational conditions that have been proposed for thermochemical energy storage applications: i) calcination in inert atmosphere and carbonation in CO 2 . ii) closed CO 2 loop and iii) closed CO 2 loop with calcination at reduced CO 2 pressure. The morphology of the prepared architectures was meticulously characterized through Xray tomography and scanning electron microscopy (SEM). The key parameters of the cast samples are comprehensively assessed to provide a thorough understanding of their properties and the impact of these multiscale structures on the reaction kinetics, conversion and energy storage performance. 2. Experimental section 2.1. Materials Limestone from the standard Eskal 500 series was obtained from KSL Staubtechnik GmbH (Germany). Samples featuring a well-defined particle size were used: 5 μ m (Dv(50) =4.71). Fig. 1 includes the particle size distribution and a digital photograph of the architectures' cross-section. ZrO 2 (Tosoh, Ref.TZ-0) and SiO 2 (99.5 % Strem Chemicals, Ref.931436) were utilized as stabilizers. The binder used in the freeze-casting process was prepared by dissolving polyvinyl alcohol (PVA, Ref.8148941001) in distilled water. 2.2. Fabrication of porous CaCO 3 architectures by freeze-casting A solution containing 40 wt% solid loading was vigorously mixed into a 4 wt% PVA solution prepared using distilled water in a sonicator bath for 10 min. For the monoliths incorporating additives, ZrO 2 and SiO 2 were manually mixed with the CaCO 3 powder for 20 min, using an agate mortar, prior to the dispersion in the sonication bath. In both cases, the additive amount to 10 wt% of the total mass of the sample. It is expected the sonication would produce a homogeneous mixture of the CaCO 3 and the additive. The prepared mixture was transferred into an acrylic glass mold with an external cylindrical diameter of 25 mm, an internal diameter of 5 mm, and a length of 15 mm. This mold was placed on the upper section of a copper rod immersed in liquid nitrogen to initiate the cooling process. The freeze-casting system ensured unidirectional solidification by cooling the copper rod from room temperature to −50 ◦C through immersion in the liquid nitrogen bath. Water removal from the samples was achieved using a freeze-dryer (Christ Gamma 2–20, Martin Christ Gefriertrocknungsanlagen GmbH, Germany) equipped with a vacuum pump. This process was conducted at −30 ◦C under a vacuum pressure of 0.03 mbar for 24 h. After sublimation, the green bodies were carefully extracted from the acrylic glass mold. As shown in Fig. 2, the process of constructing the architectures involves stirring the solution to ensure homogeneity, followed by controlled freezing to promote channel formation. During freeze drying, water is removed, creating a porous structure that enhances gas interaction with the limestone. This channel structure is preserved throughout the process, resulting in a highly porous architecture. The samples prepared by this method are named FC-CaCO 3 , FC-CaZr10 and FC-CaSi10 depending on the additive included. Prior to the multicycle tests, the binder was removed by gradually heating the samples in an air atmosphere at a rate of 1 ◦C/min until reaching 585 ◦C. 2.3. Characterization methods The morphological characteristics of the samples were analyzed using a scanning electron microscope (SEM) Hitachi S4800. Prior to imaging, the samples were sputter-coated with a thin gold layer using an Emitech K550 Telstar sputter-coating system. The 3D microstructure of the samples was analyzed using highresolution transmission X-ray computed tomography (XCT) with an Xradia Versa 610 XRM system (Zeiss), capable of achieving submicrometer resolution. A volume of approximately 300 μ m ×300 μ m × 400 μ m aligned so that the z-axis matched the freezing direction was reconstructed using a CCD camara with a 20×microscope objective coupled to the scintillator at a voxel size of 460 nm. Imaging was conducted at an acceleration voltage of 140 kV and a current of 150 μ A, ensuring adequate contrast. Image segmentation to separate the solid and pore phases was carried out using a machine learning pixel classifier [40], and histograms of solid and pore thicknesses (as a proxy to the pore size distribution) were calculated using the BoneJ plugin in FIJI [41,42]. The anisotropic tortuosity of the pore space was determined numerically by solving the diffusion equation using a finite-difference scheme [43], with calculations run on the Hercules supercomputer at the Andalusian Scientific Computing Center (CICA) – more details are given in the Supplementary information. The particle size distribution was determined through laser N. Amghar et al. Journal of Energy Storage 125 (2025) 116681 2 diffraction with a Mastersizer 2000 (Malvern). To prevent aggregation, the samples were sonicated for 30 min and dispersed in distilled water. 2.4. Multicycle experiments Multicycle experiments were conducted using a custom-built thermogravimetric reactor designed to operate across a broad pressure range (0.01 to 5 bar). The reactor included a microbalance and a Watlow furnace with a sealed, non-porous mullite tube utilizing O-rings. A flat alumina crucible, suspended from the microbalance wire, ensured effective contact between the sample and the gas. Further details of the experimental setup can be found in [21,44]. The performance of the samples in multicycle experiments was evaluated under various operating conditions, as detailed in Table 2. After binder removal, the temperature was increased to the target value at a rate of 1 ◦C/min, following the specific experimental conditions outlined in Table 1. Upon reaching the desired temperature, carbonation stages were maintained for 15 min, while calcination stages lasted 25 min. Under T1 conditions, gas changes were performed manually. For T2 conditions, the cycles were controlled by altering the temperature at a heating and cooling rate of 10 ◦C/min. In T3 conditions, the carbonation and calcination stages were regulated by adjusting the chamber pressure using a vacuum pump. 2.5. Effective conversion and residual effective conversion To evaluate the multicycle performance of the samples, the effective conversion (Eq. (2)) was employed [11]. The effective conversion is defined as the mass ratio of calcium oxide converted during the carbonation stage at each Nth cycle to the total mass of the sample before carbonation (m), including solids inert to carbonation. Therefore, the term m carb −m represents the CO 2 uptake in each cycle. The FCCaZr10 and FC-CaSi10 composites including 10 wt% of inert additive will exhibit a maximum attainable effective conversion of 0.9. 500 µm 0.1 1 10 100 0 1 2 3 4 5 6 7 8 9 Volumetric fraction (%) Particle diameter (µm) Fig. 1. Particle size distribution and a photograph of the initial architecture. Controlled freezing process Freeze drying Slurry preparation by stirring PVA chains CaCO3 powder Destilled water Liquid N2 container Methacrylate box Copper rod Acrilic glass mold Slurry Fig. 2. Schematic diagram illustrating the preparation process of calcium-derived freeze-cast materials. Table 1 Operating conditions for calcination/carbonation cycles used in this work. Test Calcination Carbonation Temperature, (◦C) Gas Absolute CO 2 pressure (bar) Temperature, (◦C) Gas Absolute CO 2 pressure (bar) T1 800 N 2 –800 CO 2 1 T2 950 CO 2 1 850 CO 2 1 T3 780 CO 2 0.1 780 CO 2 1 N. Amghar et al. Journal of Energy Storage 125 (2025) 116681 3 Xeff =mcarb −m m⋅WCaO WCO2 =mCO2(t) m WCaO WCO2 (2) In this equation, m carb is the sample mass after carbonation at the Nth cycle, m is the mass of the calcined sample at the first cycle, and W CaO (56 g/mol) and W CO2 (44 g/mol) are the molar masses of CaO and CO 2 , respectively. To compare the performance of different samples, the multicycle conversion data can generally be well fitted using the semiempirical equation [45]: Xeff,N=Xr+Xeff,1 k(N−1) + (1−Xr/Xeff,1)−1(3) In this expression, X eff,1 is the effective conversion at the first cycle, N represents the cycle number, k is the deactivation rate constant, and X r is the residual conversion towards which the conversion converges after a large number of cycles. 3. Results and discussion 3.1. Multicycle performance of the freeze-cast architectures The multicycle performance of the prepared pellets was studied according to three different operational conditions, as detailed in Table 1. Condition T1 involves calcination in N 2 and carbonation in CO 2 . This scheme is probably the most commonly used in works dealing with Calcium Looping for TCES applications [24,27,46]. In inert gas, the calcination temperature is maintained at about 750 to 800 ◦C. The carbonation reaction is carried out at about 800–850 ◦C in CO 2 to ensure a fast reaction rate [12,46]. However, this approach requires the implementation of a complex and costly gas separation stage. To address this issue, closed CO 2 loop operational schemes were devised [11,14,27]. The main drawback of such schemes is that the effective calcination temperature increases up to 950 ◦C, what leads to intense deactivation of the CaO due to sintering [11,17]. These operational conditions are represented by T2. Recently, the concept of closed CO 2 loop was extended to contemplate calcination at reduced pressures [21,47,48]. This adjustment notably lowered the effective calcination temperature and improved the residual conversion exhibited by the CaO. These operational concept is simulated by T3 conditions. Fig. 3 presents the temperature and mass time evolutions measured during a test carried out under T1 conditions for the FC-CaCO 3 sample, along details of the second and tenth cycles. Before starting the cycle process, the binder used in the manufacturing of the monoliths was debinded in air at 585 ◦C. Due to this thermal treatment, PVA chains are volatilized, providing a porous structure and promoting an initial mass loss due to the degradation of the binder. In the literature, it is reported that PVA is fully decomposed at temperatures below 500 ◦C with residues close to zero [49]. In this study, approximately 2.4 % of PVA was incorporated into the monoliths to facilitate gel formation and ensure proper dispersion of CaCO 3 . This percentage correlates well with the mass loss observed during the debinding stage. Following the binder degradation, the Calcium Looping (CaL) cycles 0 250 500 750 1000 1250 1500 1750 0 100 200 300 400 500 600 700 800 Time (min) Temperature (ºC) 55 60 65 70 75 80 85 90 95 100 Mass (%) Air debinding N2/CO2 calcium looping a) CO2N2CO2N2 2410 2420 2430 2440 2450 2460 Time (min) 60 70 80 90 100 Mass (%) 1170 1180 1190 1200 1210 Time (min) 60 70 80 90 100 Mass (%) b) c) Fig. 3. Time evolution of temperature and sample mass for the limestone-derived freeze-cast materials during multicycle tests under T1 conditions: a) 20 cycles, b) detail of the second cycle and c) detail of the tenth cycle. N. Amghar et al. Journal of Energy Storage 125 (2025) 116681 4 commence. As illustrated in Fig. 3b and c, carbonation proceeds rapidly as soon as CO 2 enters the chamber [12]. Under the experimental conditions used, the cycle exhibits a characteristic profile, including a slowdiffusion stage that follows the initial fast carbonation. In powder form, carbonation occurs swiftly at the surface of the CaO particles through a reaction-controlled mechanism [50,51]. However, the formation of a CaCO 3 layer on the particle surface during this stage obstructs CO 2 diffusion to the unreacted CaO core. Consequently, the reaction progresses via solid-state diffusion through the CaCO 3 blocking layer [52–54]. In contrast, the calcination stage is initiated with a delay after the introduction of nitrogen, as certain time is required to purge residual CO 2 from the chamber to values below the equilibrium pressure at the reaction temperature. This process is inherently slower, with parameters such as the gas flow rate, CO 2 diffusivity in the purge gas, reactor design and particle sizes significantly influencing the reaction kinetics. Lower purge flow rates extend the time required for CO 2 to diffuse outside the particles, particularly in the case of large particles for where internal mass transfer resistances become a limiting factor [55,56]. Fig. 4 compares the multicycle performance of various samples in terms of CaO conversion, as determined through thermogravimetric experiments and calculated using Eq. (2). Table 2 presents a summary of both the initial and the residual conversion values derived from mathematical fittings to Eq. (3). It is well established that the reactivity of CaO progressively declines across successive carbonation and calcination cycles. This decline in CaO conversion is primarily due to extensive sintering caused by the high temperatures required for the process, which leads to a substantial loss of surface area [16,57,58]. Furthermore, under the typical operation conditions for TCES -which involve calcination in an inert gas environment followed by high-temperature carbonation in CO 2 - pore plugging becomes a limiting factor [58,59]. Pore plugging occurs when the surface porosity of the CaO particles is obstructed by the rapidly forming CaCO 3 layer. This layer hinders CO 2 diffusion, thereby restricting the maximum conversion that can be achieved during each carbonation phase. Additionally, large particles are more prone to pore-plugging phenomena [58,60]. It is anticipated that the freeze-cast structures constructed using small particles can enhance multicycle performance by mitigating both sintering-induced deactivation and pore blockage issues. This hypothesis is supported by the multicycle experiments conducted in this study. For samples tested under N 2 /CO 2 atmospheres (Fig. 4a), the activity of FC-CaCO 3 was observed to decrease from 0.72 in the first cycle to a residual value of 0.56. In contrast, the multicycle performance of CaCO 3 powder tested under similar conditions exhibit much smaller residual conversion: 0.3. This constitutes a significant improvement. The considerable variability in operating conditions, particle sizes and equipment used in the literature complicates direct comparison with previous results. Nevertheless, the observed residual conversion value is comparable to, or even exceeds, those reported for CaO prepared by wet chemical methods, which typically produce particle sizes smaller than those employed in this work [23,24,61]. Even more notable is the stabilization of the conversion from the tenth cycle onward. Previous studies with small CaO particles typically report continuously decreasing trends due to the high sinterability of such small particles [24,29,62,63]. This suggests that the freeze-cast structures are less prone to sintering-induced deactivation, probably due to limited inter-particle contacts compared to loose powder. The T1 conditions were also employed to assess the performance of FC-CaZr10 and FC-CaSi10 samples, which were prepared through casting with the addition of ZrO 2 and SiO 2 , two widely employed structural stabilizers. Under typical CaL operating conditions, SiO 2 readily forms calcium silicates that act as a supporting skeleton that inhibit the sintering of CaO particles [25,64]. ZrO 2 serves a similar purpose; however, its higher thermal stability makes the calcium zirconates to appear in a longer timeframe and require higher temperatures [11,23,63,65]. X-ray diffraction patterns of the cycled specimens are provided in Fig. S1. Interestingly, the incorporation of these additives to the freeze-cast structures did not enhance the multicycle behavior of CaO. In fact, the FC-CaCO 3 samples demonstrated significantly better performance than FC-CaZr10 and FC-CaSi10. The notable reduction in conversion shown by FC-CaSi10 samples, with an estimated residual conversion of 0.34, can be attributed to the removal of active CaO due to the formation of calcium silicates. Alternatively, no significant amount of calcium zirconates are expected under T1 conditions. Thus, the residual conversion observed in FC-CaZr10 is 0.51, only slightly smaller than FC-CaCO 3 . The limited reduction in conversion can be attributed to the presence of inert ZrO 2 , what reduces the maximum effective conversion theoretically 010203040 0.0 0.2 0.4 0.6 0.8 1.0 010203040 0.0 0.2 0.4 0.6 0.8 1.0 Raw CaCO3-T1 FC-CaCO3-T1 FC-CaZr10-T1 FC-CaSi10-T1 Xeff N a) b) FC-CaCO3-T1 FC-CaCO3-T2 FC-CaCO3-T3 Xeff N Fig. 4. Comparison of the evolution of the multicycle conversion obtained for: a) samples subjected to T1 conditions, and b) FC-CaCO 3 samples subjected to T1, T2 and T3 operating conditions. Table 2 Main results obtained from fitting of the conversion results. Effective conversion at cycle 1 (X eff1 ), residual conversion (X r ), deactivation constant (k) and Rsquared (R 2 ). Sample X eff,1 X r k R 2 Raw CaCO 3 -T1 0.77 0.30 1.360 0.832 FC-CaCO 3 -T1 0.72 0.56 3.471 0.875 FC-CaZr10-T1 0.65 0.51 4.071 0.867 FC-CaSi10-T1 0.59 0.34 0.637 0.991 FC-CaCO 3 -T2 0.83 0.16 0.472 0.993 FC-CaCO 3 -T3 0.94 0.69 1.183 0.985 N. Amghar et al. Journal of Energy Storage 125 (2025) 116681 5 attainable. This suggests that the porous architecture of the freeze-cast materials results in inherently stable architectures that renders the use of stabilizing additives unnecessary, as its contribution as sintering inhibitors is played by the open freeze-cast architecture. Similar stable behaviors have been reported in hollow microspheres or core-shell morphologies fabricated through template synthesis, where the available free space accommodates volume changes and mitigates sinteringinduced deactivation [28,66,67]. However, in those cases, metal oxides remain essential to preserve the structural integrity of the porous structure. In contrast, freeze cast architectures do not seem to require the inclusion of structural stabilizing additives. Fig. 4b shows the multicycle performance obtained in tests conducted under CO 2 closed-loop conditions. This operational scheme helps to avoid the high cost and added complexity associated with gas separation membranes [14,68]. However, operating at atmospheric pressure under CO 2 (T2 conditions) requires higher calcination and carbonation temperatures, which leads to accelerated sintering and rapid deactivation of the limestone [11,17]. As a result, FC-CaCO 3 samples tested under T2 conditions exhibit a much lower residual conversion of 0.16. However, this residual conversion is still greater than the values attained for CaCO 3 powder under similar conditions in the same apparatus [17]. Recent studies have demonstrated that operating under mild vacuum conditions can mitigate the deactivation caused by sintering [47]. Working at low pressure shifts the thermodynamic equilibrium of the CaCO 3 /CaO system to a lower temperature, thereby facilitating calcination at reduced temperatures. Thus, T3 conditions involve isothermal calcination at 780 ◦C under 0.1 bar absolute pressure of CO 2 and carbonation at 1 bar of CO 2 . The FC-CaCO 3 sample subjected to T3 conditions demonstrates the best performance, with a very high residual conversion of 0.69. Again, such residual value constitutes a significant improvement over the value obtained in a previous work for CaCO 3 powder under similar conditions in the same instrument [47]. Nevertheless, the comparison should be taken with care as the particle size used in the previous work was much larger. These results again demonstrate the improved carbonation performance attained by CaO materials under reduced pressure conditions. While operation under vacuum implies an additional investment cost and extra power consumption, low pressure (vacuum) facilities are usual at industry level. For instance, in steam power plants, in fertilizers industry and in petroleum-related activities. Such experience could be extrapolated to facilities for energy storage with Calcium Looping. 3.2. Carbonation and calcination kinetics in the freeze-cast CaO architectures Fig. 5 illustrates the kinetic behavior of the calcination and carbonation processes for the most representative samples analyzed in this study (FC-CaCO 3 -T1 and FC-CaCO 3 -T3), focusing on the second and tenth cycles as benchmarks of performance. As discussed in the previous section, CaO carbonation obeys a two-phase process: an initial rapid reaction phase followed by a significantly slower diffusion-controlled stage [19,51]. The contrast between both stages is evidenced in Fig. 5. During the reaction-controlled phase, carbonation occurs on the exposed surface of CaO, leading to the formation of dispersed CaCO 3 islands that eventually coalesce into a continuous CaCO 3 layer on the surface of the CaO particles [69,70]. Thus, the extension of the carbonation during this stage depends on the external diffusion of gas molecules to the sorbent's surface and then the intraparticle and interparticle diffusion through the porosity of the sorbent. As a result, it depends on the reaction temperature [13], partial pressure of CO 2 [12], particle size [60] and porosity [27]. Once the surface CaCO 3 forms, the reaction proceeds through a much slower solid-state diffusion mechanism [52]. Higher diffusive FC-CaCO3-T1 FC-CaCO3-T3 a) b) d)c) 0 2 4 6 8 1012141618202224 -0.22 -0.20 -0.18 -0.16 -0.14 -0.12 -0.10 -0.08 -0.06 -0.04 -0.02 0.00 dXeff/dtime Time (min) 012345678910 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Xeff Time (min) FC-CaCO3-T1 FC-CaCO3-T3 FC-CaCO3-T1 FC-CaCO3-T3 012345678910 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 dXeff/dtime Time (min) 024681012141618202224 Xeff Time (min) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Fig. 5. Time evolution of CaL process: a) conversion evolution for the carbonation process b) conversion evolution for the calcination process c) first derivative of the effective conversion for the carbonation process and d) first derivative of the effective conversion for the calcination process. Continuous line indicates results from the second cycle and the dashed line from the tenth cycle. N. Amghar et al. Journal of Energy Storage 125 (2025) 116681 6 carbonation rates are observed in milled samples [70], materials with larger pores [62], small particles [60], steam [22] and additives that promote the diffusion of CO 3 2− such as molten carbonates [71], or sintering inhibitor additives [11]. FC-CaCO 3 -T3 demonstrates greater reactivity in this phase than FCCaCO 3 -T1, both in the second and tenth cycle. The diffusion controlled stage for both FC-CaCO 3 -T3 and FC-CaCO 3 -T1 exhibit the seemingly flat profile typical of reaction schemes comprising calcination in inert atmosphere and carbonation at high temperatures in CO 2 [60]. However, FC-CaCO 3 -T3 achieves slightly faster carbonation rate, evidenced by the slope, what can produce relevant differences at longer carbonation times. Since the starting morphology of the sample is similar, the enhanced performance can only be attributed to the milder calcination conditions used in T3, which help mitigate deactivation by sintering. Furthermore, studies have shown that calcination at reduced pressure produces CaO with very small crystallite size and a larger surface area [21,47]. The fact that the carbonation rate remains largely unchanged between the 2nd to the 10th carbonation cycles suggests that the favorable pore structure achieved through the freeze-cast process is better preserved in T3. In contrast, under T1 conditions, the carbonation rate at the 10th cycle is slightly slower than at the 2nd cycle, indicating a gradual evolution in the pore structure. The calcination reaction is the slowest process, requiring 8–12 min for completion, depending on the conditions. The calcination rate of the structures is also faster under T3 conditions. 3.3. Morphology study for the freeze-casting monoliths The study of morphology is critical for understanding the behavior of the sample in terms of conversion extension and reaction kinetics process, as efficient gas flow within the sample and pore blockage have significant influence. In this context, the evolution of the conversion along the cycles as shown in Fig. 4 are consequence of morphological changes. Fig. 6a and b display SEM images of the FC-CaCO 3 -T1 at cycle 1. The material appears composed of particles loosely bonded together. Particle size is in agreement with the distribution shown in Fig. 1. The contours of the individual particles are evident and the degree of sintering is still limited. The observed porosity is attributed to the freezecasting fabrication method and binder decomposition during the initial processing stage. After 40 cycles (Fig. 6c and d), noticeable changes are evident in the high-magnification images, revealing a significant increase in particles size. This change is a consequence of the sintering process, which leads to a decrease of surface nano-sized porosity and an increase of interparticle pore volume [72]. Despite the large number of cycles, individual particles remain distinguishable, ensuring good conversion rates and cycle stability throughout the cycling process. According to the literature, limestone subjected to 20–30 thermal cycles typically exhibits a more extensive sintering, with clear merging of neighboring grains [24]. XCT observations of the samples subjected to one and 40 carbonation cycles confirm the SEM observations and provide insight into the microstructural evolution during cycling. Fig. 7 summarizes these observations using two-dimensional sections for ease of visualization (three-dimensional volumes are shown in the Supplementary video). Fig. 7a shows the original microstructure of freeze-cast samples, exhibiting anisotropic porosity with elongated channels along the freezing direction. Over the cycles, this geometry changes, resulting in an isotropic distribution of the pores along the observed surface, as shown in Fig. 7d for the sample with 40 cycles. These results are consistent with the tortuosity measurements presented in the supporting information, where the value obtained in the z-direction decreases over the CaL cycles. The changes observed between cycles 1 and 40 could be related to the preferential shrinkage of freeze-cast samples in the freeze casting direction, which reduces pore orientation [73]. The pore size and particle size distributions were quantified using a thickness algorithm. This algorithm calculates, for each point in the pore or solid space, the diameter of the largest sphere that fits the pore or solid and contains the point, serving as a proxy for determining the pore and particle size distributions. Results of these calculations are shown in Fig. 7b and c for the pores and solid particles, respectively, for the sample subjected to 1 cycle. After 40 cycles, the sample exhibits a clear increase in pore size, likely due to the sintering process occurring during the cycles at high temperatures. Calculated pore and particle size distributions are presented in Fig. 8, which were fitted to log-normal functions to determine the average pore and particle sizes. Comparison of cycle 1 to cycle 40 shows an opening of porosity, most likely due to local sintering effects in the solid walls. This results in increased pore volume fraction (from 55 % vol to 63 % vol) and average pore size (from 5.0 μ m to 9.0 μ m), as well as a broadening of the pore size distribution. The mean particle size also increases slightly upon cycling, from 4.7 μ m 10 µm b) 10 µm d) a) 2 µm c) 2 µm Fig. 6. Morphology of the carbonated FC-CaCO 3 -T1 monolith at ×5000 and ×1000 magnification: a) cycle 1 at high magnification, b) cycle 1 at low magnification, c) cycle 40 at high magnification and d) cycle 40 at low magnification. N. Amghar et al. Journal of Energy Storage 125 (2025) 116681 7 to 6.0 μ m, which is consistent with the effect of local sintering. An increase in pore size due to sintering processes has been reported by other authors in structures prepared by freeze casting in ceramic materials [74]. Changes in particle size were also observed using XCT in ceramic samples prepared by freeze casting [75]. This morphological evolution may explain the changes in the carbonation rate between the 2nd and 10th cycles, as observed in Fig. 5c. It is important to note that due to the limited resolution of XCT compared to SEM, what we refer to as particle size in this context is not the primary particle or crystallite size, but rather particle agglomerates of a few μ m and larger, such as those shown in Fig. 6, which indeed shows qualitatively the same agglomerate enlargement upon cycling. The calculated tortuosity along the freezing direction is τ 1 =2.0 for cycle 1 and τ 2 =1.5 for cycle 40, which can be rationalized in terms of an a) b) c) d) e) f) 50 µm 50 µm 50 µm 50 µm 50 µm 50 µm 25 µm 20 µm 15 µm 10 µm 5 µm 0 µm 25 µm 20 µm 15 µm 10 µm 5 µm 0 µm 25 µm 20 µm 15 µm 10 µm 5 µm 0 µm 25 µm 20 µm 15 µm 10 µm 5 µm 0 µm Fig. 7. Microstructural evaluation of FC-CaCO 3 -T1 monoliths by high resolution XCT: a) absorption contrast section after 1 cycle, b) pore size quantification after 1 cycle, c) solid phase quantification after 1 cycle, d) absorption contrast section after 40 cycles, e) pore size quantification after 40 cycles, c) solid phase quantification after 40 cycles. Each voxel is colored according to the size of the largest sphere that fits the pore space and contains the voxel. In all images, the freezing front direction is vertical in the page. 1 cycle 40 cycles 0 5 10 15 20 25 0 5 10 15 a) b) Volume fraction (%) Pore diameter (µm) 55% pore volume mean pore d= (5.0 ± 1.7) µm Pore diameter (µm) 0 5 10 15 20 25 0 5 10 15 Volume fraction (%) 63% pore volume mean pore d= (9 ± 4) µm 0 5 10 15 20 25 0 5 10 15 45% solid volume mean particle d= (4.7 ± 1.2) µm Volume fraction (%) Particle diameter (µm) Particle diameter (µm) 0 5 10 15 20 25 0 5 10 15 37% solid volume mean particle d= (6.0 ± 1.2) µm Volume fraction (%) c) d) Fig. 8. Distribution of the pore and solid fraction of the FC-CaCO 3 -T1 monoliths: a) pore size and distributions after 1 cycle, b) pore size and distributions after 40 cycles, c) solid particle size and distributions after 1 cycle and d) solid particle size and distributions after 40 cycles. N. Amghar et al. Journal of Energy Storage 125 (2025) 116681 8 increase of microporosity and opening of the pore space, which facilitates gas flow inside the porous monoliths. Both values are quite low for macroporous structures, confirming that the special microstructure obtained by freeze-casting contributes to fast gas diffusion and helps explain the excellent observed kinetics. 4. Conclusions This study validates the feasibility and effectiveness of freeze-casting for fabricating porous calcium-based monoliths tailored to CaL in thermochemical energy storage. By employing polyvinyl alcohol as a binder, the method yielded highly porous scaffolds through controlled freezing in cylindrical molds and subsequent freeze-drying. Multi-cycle tests demonstrated outstanding performance, with the FC-CaCO 3 -T3 monolith achieving a notable residual conversion of 0.69 under moderate vacuum calcination, surpassing conventional powder-based systems. Monoliths calcined in inert atmospheres also attained a residual conversion of 0.56, underscoring the advantages of operating at reduced CO 2 pressure, which involves lower calcination temperatures, reduced deactivation, and shorter calcination/carbonation times. SEM and XCT analyses confirmed the formation of a robust porous network that maintained reactive surfaces and mitigated sintering over multiple cycles. Although additives were introduced, they did not enhance performance, affirming the inherent structural stability of the monoliths. While high-temperature exposure induced modest increases in pore volume and particle size, these changes did not compromise overall efficacy. Collectively, these findings position freeze-casting as a versatile, scalable approach for developing advanced materials capable of meeting industrial demands for durable, high-temperature energy storage. Supplementary data to this article can be found online at https://doi. org/10.1016/j.est.2025.116681. CRediT authorship contribution statement Nabil Amghar: Investigation, Formal analysis. Juan Ivorra-Martinez: Writing – original draft, Formal analysis, Data curation. Antonio Perej´ on: Supervision, Methodology, Conceptualization. Dorian Hanaor: Supervision, Methodology. Aleksander Gurlo: Resources. Joaquín Ramírez-Rico: Validation, Methodology, Formal analysis. Luis A. P´ erez-Maqueda: Visualization, Resources, Project administration, Methodology, Conceptualization. Pedro E. S´ anchez-Jimenez: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments Financial support is acknowledged from grants TED2021-131839BC22 and PDC2021-121552-C21 (MCIN/AEI/10.13039/501100011033 and European Union Next Generation EU/PRTR) and by European Union NextGenerationEU/PRTR, and the grant PID2022-140815OBC22 funded by MCIN/AEI/10.13039/501100011033 and ERDF A way of making Europe. PhD Fellowship granted to N. Amghar (PRE2018085866) from the Spanish Government Agency Ministerio de Ciencia, Innovaci´ on y Universidades. J. Ivorra-Martinez thanks Generalitat Valenciana - GVA for funding a postdoc position through the CIAPOS program co-funded by ESF Investing in your future, grant number CIAPOS/2023/362. XCT measurements were performed at the X-ray laboratory of the Research, Technology and Innovation Center (CITIUS) of the University of Seville. Data availability Data will be made available on request. References [1] Y. Zhang, et al., Calcium looping for CO 2 capture and thermochemical heat storage, a potential technology for carbon neutrality: a review, Green Energy and Resources 2 (3) (2024) 100078. [2] U. Pelay, et al., Thermal energy storage systems for concentrated solar power plants, Renew. Sustain. Energy Rev. 79 (2017) 82–100. [3] R. Chacartegui, et al., Thermochemical energy storage of concentrated solar power by integration of the calcium looping process and a CO 2 power cycle, Appl. Energy 173 (Supplement C) (2016) 589–605. [4] J. Chen, L. Duan, Z. Sun, Review on the development of sorbents for calcium looping, Energy Fuels 34 (7) (2020) 7806–7836. [5] W. Li, et al., Study on CaO-based materials derived from steel slag for solar-driven thermochemical energy storage, Sol. Energy Mater. Sol. Cells 277 (2024) 113093. [6] M. Imani, M. Tahmasebpoor, P.E. S´ anchez-Jim´ enez, Metalbased eggshell particles prepared via successive incipient wetness impregnation method as a promoted sorbent for CO 2 capturing in the calcium looping process, J. Environ. Chem. Eng. 11 (5) (2023) 110584. [7] H. Wu, et al., Review of solar thermochemical heat storage equipment and systems based on calcium-looping, Journal of Energy Storage 103 (2024) 114146. [8] D. Rodrigues, et al., Optimization of an improved calcium-looping process for thermochemical energy storage in concentrating solar power plants, Journal of Energy Storage 72 (2023) 108199. [9] C. Ortiz, et al., Process integration of calcium-looping thermochemical energy storage system in concentrating solar power plants, Energy 155 (2018) 535–551. [10] S. Pascual, L.M. Romeo, P. Lisbona, Optimized Ca-looping thermochemical energy storage under dynamic operation for concentrated solar power, Journal of Energy Storage 68 (2023) 107587. [11] B. Sarri´ on, et al., Role of calcium looping conditions on the performance of natural and synthetic Ca-based materials for energy storage, Journal of CO2 Utilization 28 (2018) 374–384. [12] J. Arcenegui-Troya, et al., Overlooked pitfalls in CaO carbonation kinetics studies nearby equilibrium: instrumental effects on calculated kinetic rate constants, Alex. Eng. J. 61 (8) (2022) 6129–6138. [13] Y.A. Criado, B. Arias, J.C. Abanades, Effect of the carbonation temperature on the CO 2 carrying capacity of CaO, Industrial & Engineering Chemistry Research 57 (37) (2018) 12595–12599. [14] A. Alovisio, et al., Optimizing the CSP-calcium looping integration for thermochemical energy storage, Energ. Conver. Manage. 136 (2017) 85–98. [15] C. Ortiz, et al., Carbonation of limestone derived CaO for thermochemical energy storage: from kinetics to process integration in concentrating solar plants, ACS Sustain. Chem. Eng. 6 (5) (2018) 6404–6417. [16] A.A. Scaltsoyiannes, A.A. Lemonidou, On the factors affecting the deactivation of limestone under calcium looping conditions: a new comprehensive model, Chem. Eng. Sci. 243 (2021) 116797. [17] B. Sarrion, et al., Pressure effect on the multicycle activity of natural carbonates and a Ca/Zr composite for energy storage of concentrated solar power, ACS Sustainable Chemistry and Engineering 6 (6) (2018) 7849–7858. [18] R.H. Borgwardt, Calcium oxide sintering in atmospheres containing water and carbon dioxide, Ind. Eng. Chem. Res. 28 (4) (1989) 493–500. [19] P.E. S´ anchez-Jim´ enez, et al., Influence of ball milling on CaO crystal growth during limestone and dolomite calcination: effect on CO2 capture at calcium looping conditions, Crystal Growth and Design 16 (12) (2016) 7025–7036. [20] R. Han, et al., Progress in reducing calcination reaction temperature of calciumlooping CO 2 capture technology: a critical review, Chem. Eng. J. 450 (2022) 137952. [21] N. Amghar, et al., Thermochemical energy storage using calcium magnesium acetates under low CO 2 pressure conditions, Journal of Energy Storage 63 (2023) 106958. [22] J. Arcenegui-Troya, et al., Steam-enhanced calcium-looping performance of limestone for thermochemical energy storage: the role of particle size, Journal of Energy Storage 51 (2022) 104305. [23] A. Antzara, E. Heracleous, A.A. Lemonidou, Improving the stability of synthetic CaO-based CO 2 sorbents by structural promoters, Appl. Energy 156 (2015) 331–343. [24] H. Zhang, et al., Study of co-promotion mechanism of Zr-Mn co-doped CaO-based composite for solar thermochemical energy storage, Journal of Energy Storage 97 (2024) 112877. [25] P.E. Sanchez-Jimenez, L.A. Perez-Maqueda, J.M. Valverde, Nanosilica supported CaO: a regenerable and mechanically hard CO 2 sorbent at Ca-looping conditions, Appl. Energy 118 (2014) 92–99. [26] C. Huang, et al., Hierarchically porous calcium-based composites synthesized by eggshell membrane templating for thermochemical energy storage of concentrated solar power, Journal of Energy Storage 52 (2022) 104769. [27] X. Huang, et al., Preparation of morph-genetic aluminum-doped calcium oxide templated from cotton and the calcium looping performance for energy storage in the presence of steam, Journal of Energy Storage 72 (2023) 108325. [28] A. Kurlov, et al., CaO-based CO 2 sorbents with a hierarchical porous structure made via microfluidic droplet templating, Ind. Eng. Chem. Res. 59 (15) (2020) 7182–7188. N. Amghar et al. Journal of Energy Storage 125 (2025) 116681 9