Fluidisation of Thermochemical Energy Storage Materials: Degradation Assessment

ABSTRACT Composite zeolites impregnated with anhydrous salt particles are promising materials for use in domestic thermochemical energy storage (TCES), however they have limited power output. Fluidization has the potential to increase the mass and heat transfer of TCES systems. In this study, a composite TCES material of LiX zeolite impregnated with MgCl2 salt is produced. This material and a LiX zeolite batch are then fluidized at different gas velocities to study whether degradation of these particles occurs when fluidizing. Images of the particles after fluidizing are analyzed using a validated method utilizing microscope images and specialized software used to study whether fragmentation or abrasion have been induced by fluidization. Composite particles were found to have increased average diameter by 7%−8%, which could be due to imperfect dehydration of these particles after the impregnation process, but also due to observed salt conglomerates present on the surface of some particles. Overall, excess degradation was not found due to fluidization at any gas velocity. No mass loss was measured in any sample after fluidization, and no appreciable fragmentation was observed. The LiX zeolite and composite particles measured 0.6192 mm and 0.6799 mm prior to fluidization and measured 0.6405 mm and 0.7088 mm after fluidization. However, this change is not statistically significant, and it is shown that a change in diameter is unlikely. Despite this, some fines have been produced which were found to have an average diameter of 3.60 μm (σ = 1, ±2.34 μm), and as such present a hazard if inhaled.


Introduction
Thermal energy storage (TES) is one of the key technologies required to shift our energy systems toward a more sustainable future.The key characteristics of TES are energy density, charge and discharge rates, and storage time (Zhang et al. 2016).Energy density is important due to the need for large quantities of energy to be stored in practical volumes for many applications, especially when storing energy for a seasonal deficit and for domestic use.Charge and discharge time are key for two reasons: The first reason is that TES can utilize a quick charge time in order to effectively store the maximum available energy in some applications.In the case of domestic solar energy, the power output is unlikely to be high; however, in industrial processes the energy may only be available at high powers, and as such a fast charge time is required to store the maximum available energy (Wünsch et al. 2020).In order to maximize the levelized cost of energy (LCOE) of a plant, cycle power must be taken into consideration (González-Roubaud, Pérez-Osorio, and Prieto 2017).The second reason is to meet the power demand required from a TES for a particular application; for domestic applications a TES must meet the power required by hot water demands, for example (approximately 15 kW, at a temperature of approximately 60°C for a typical hot shower (Ma, Bao, and Roskilly 2018)).The same applies for industrial applications, the power output must match the demand, hence discharge time is a key characteristic (Alva, Lin, and Fang 2018).Storage time is also highly important so that TES is able to be applied to the issue of seasonal and long-term TES as well as diurnal and short-term TES (Bott, Dressel, and Bayer 2019).To mitigate the seasonal deficit in energy, TES systems must be able to store energy on the order of months (Ma, Bao, and Roskilly 2018).In this field, thermochemical energy storage is of particular interest due to its ability to store thermal energy indefinitely (Kerskes 2016).
Within the field of thermochemical TES, thermochemical energy storage (TCES) is emerging as a frontrunner to solve the issue of seasonal TES (Xu, Wang, and Li 2014).The ability of TCES materials to store energy for an indefinite period puts them above TES methods which are subject to energy loss over time."Salt in porous matrix," i.e., composite TCES materials, has shown increased performance with respect to energy density over traditional TCES materials such as pure zeolites, and silica gel (Hongois et al. 2011).Numerous studies have been conducted in this field, with Hongois et al. (2011) first applying a MgSO 4 based composite TCES to seasonal heat storage.This study demonstrated an improved energy density of 166 kWh/ m 3 , compared to 136 kWh/m 3 for pure 13X zeolite.Another study of 13X zeolite and MgCl 2 , which also utilized the deliquescence reaction of the salt, demonstrated an energy density of 308 kWh/m 3 (Xu et al. 2019).This is the closest to the theoretical maximum energy density of 694.44 kWh/m 3 that can be found in the literature (Van Essen et al. 2009).The salt, which is impregnated into a porous host matrix, acts as the primary energy carrier allowing for increased energy storage potentialat the same time, the host matrix stabilizes the salt, preventing caking from occurring, as well as providing a large surface area through which adsorption (a key process in TCES) can occur (Aristov and Gordeeva 2009).The rate at which this adsorption occurs is directly proportional to the power output of the TCES system, and thus it is clear that composite TCES materials have a highly promising future in TES as a high energy density storage medium.
One of the key drawbacks of adsorption TCES is its relatively limited power output (4.4 kW being the maximum attained in a domestic scale pilot system (van Alebeek et al. 2018)), which is an issue previously highlighted (Marie et al. 2022).Current-packed bed reactors result in poor reaction kinetics (Almendros-Ibáñez et al. 2019) and slow reaction times which results in this limited power output.It has been suggested that the application of novel reactor types in this area could significantly improve these key metrics (Zondag, Kalbasenka, and Van Essen 2008).This could potentially increase the power output of adsorption TCES significantly (Marie and O'Donovan 2022).These novel reactor types include fluidized bed reactors and other moving bed reactors such as stirring and falling beds.Fluidized bed reactors have been found to have a heat transfer rate an order of magnitude higher than that of packed beds (Almendros-Ibáñez et al. 2019).As such, a fluidized bed reactor applied to domestic scale thermal energy storage could produce 40-50 kW, which would be more than enough to supply the heat needs of a household.While these reactors could potentially increase power output, they come with their own associated drawbacks, such as the parasitic energy loss from the movement of the particle bed and the high complexity associated with a fluidized bed system (Marie et al. 2022).Fluidized bed reactors, and to a lesser extent moving bed reactors, also cannot utilize the thermal stratification present in a fixed bed reactor, resulting in an exergy loss (Almendros-Ibáñez et al. 2019).There is also a potential for moving and fluidized bed reactor types to accelerate material degradation, a key issue of TCES, through the increased attrition and fragmentation of particles.A review of the literature has shown a strong dearth of study into sorption TCES in fluidized beds, with only a couple of disparate studies taking place.This is the first study attempting to model the process of adsorption in a fluidized bed (Darkwa, Ianakiev, and O'Callaghan 2006).The second study began the experimental study of the thermal performance of a fluidized TCES system (Bardy et al. 2020).Neither of these studies have investigated the degradation brought about by fluidization.An understanding of the propensity for TCES particles to degrade is of vital importance to the design of any system incorporating fluidization.Some initial studies have been conducted on the use of fluidized beds in microencapsulated-phase change materials (Keshavarz, Assari, and Basirat Tabrizi 2022).Here, it has been found that the use of pulsating fluidized beds significantly increases the storage efficiency of the medium.A reasonable proportion of study on how moving and fluidized bed affect reaction-based storage have also taken place.It has also been shown that fluidized beds are capable of increasing the efficiency of reaction TCES (Zheng et al. 2023).Mejia et al. (2022) studied how moving beds influenced the hydration of Ca(OH) 2 .This study focused on proving the concept, however the gassolid heat transfer was high, and no adverse effects were observed.Han et al. (2020) extensively reviewed the use of fluidized beds in reaction-based thermochemical energy storage.It was concluded that fluidization led to high rates of heat and mass transfer, and strong isothermization which are all characteristics highly common in fluidized beds.If these same characteristics can be translated to sorption-based TCES, then the issue of limited power outputs may be resolved.
Degradation occurs through two primary mechanisms: abrasion and fragmentation (Amblard et al. 2015).The former occurs when small particles are removed from the surface of the primary particle, producing fines.The latter occurs when the primary particle is broken apart into several small particles, destroying the primary particle.Abrasion will have the effect of producing a new small particle size distribution, as well as shifting the initial distribution toward a smaller diameter.The initial distribution will also be flattened slightly.Fragmentation results in a larger shift of the initial distribution toward a small-diameter axis, as well as flattening the distribution significantly.Another distribution of very small particles will also be produced, with a very large mass fraction.These phenomena are diagrammatically presented in Figure 1 (Pis et al. 1991).
It has been found that in some cases, composite TCES particles can experience cracking and fragmentation after a TCES cycle (Xu et al. 2019).This fragmentation is not experienced by the noncomposite counterpart of this particle, and as such is explained by the higher sorption capacity of the composite.This appears to lead to a greater change in particle diameter and as such increases the surface and internal stresses experienced by the particle.As discussed by Xu et al. (Xu et al. 2019), this does not present an issue as the sorption capacity is not significantly affected and remains close to the original sorption capacity.However, an issue may present itself if these composite particles are to be used in a novel reactor such as a fluidized bed reactor.In this setting, a change in particle size would result in a change in fluidization characteristics and present various issues.First among these issues would be the change of the hydrodynamic properties of the particles.A change in diameter would result in a fluidized bed that was no longer optimized for the correct particle diameter.But perhaps more importantly, there is a potential that the particles simply become no longer suited to fluidization and move into the region of Geldart class C particles.These are particles which are too small and do not have enough density to be fluidized (Kunii and Levenspiel 1969).Cohesive forces, such as van der Waals force, between these small particles are strong enough that fluidization cannot occur because the particles remained caked together.This results in two phenomena known as slugging and channeling, respectively, which can be seen in Figure 2.These two phenomena are both indicative of poor mixing and, as such, a low reaction rate and so are not favorable.
Another issue arising from the attrition of particles in a fluidized bed would be the potential loss of TCES material as fines are entrained in the flow and removed from the bed, this could also result in the contamination of the fluid which would be highly unfavorable in an open TCES system.Yet, no studies have been conducted which investigate the degradation of fluidized sorption-based TCES particles at differing gas velocities.As such, the study of this would be valuable and novel.If it is the case that the particles do not experience significant degradation as a result of fluidization, the cause of the previously observed fragmentation (as in Xu et al. (2019), who noted that moisture sorption performance was not affected by degradation) of the particles will be isolated to the adsorption cycling of the particles.General conclusions will be able to be drawn as to the suitability of fluidized bed reactors in adsorption based composite TCES, potentially opening the door to a new era of high-power, highenergy density, long-term TES.
The objectives of this study are as follows: Investigate whether the forces and thus stresses produced on the particles through fluidization induce a macro-level degradation of composite and noncomposite thermochemical energy storage.The second objective is to isolate whether attrition and degradation are exacerbated or induced by salt impregnation.The third objective, if there is an induced degradation, is to investigate the primary mode of degradation (abrasion as in the case of lowstress intensity or fragmentation in the case of high-stress intensity).This will demonstrate whether the particles themselves are significantly weakened in the production of the composite, so as to make composite TCES unsuited to fluidized bed reactors.The diameter distribution of the particles will be used to assess the mode of degradation (as in Figure 1).It should be noted that the focus of this study is the degradation of previously studied composite TCES particles, and so the study of the porosity and crystal structure of these particles is out within the scope of the research conducted here.

Methodology
The first step is to produce the composite particle for testing in a fluidized bed.Following this, a fluidized bed will need to be designed and constructed.The fluidized bed will need to be able to consistently fluidize Geldart class B particles at different superficial gas velocities, and it must be modular so that deconstruction and cleaning between tests is possible (as significant quantities of fines are expected to be produced).
A particle which is able to be impregnated with anhydrous salt without experiencing very adverse cracking and damage is required.The particle must also be of the correct diameter to be fluidized and be an "active matrix" meaning, it must be able to function as a TCES material by itself prior to impregnation.As such, a LiX zeolite which has been shown to fit these characteristics was selected (Touloumet et al. 2022).The zeolite was sourced from the Shanghai HengYe Chemical Industry Company.The product type was their HYGB100D molecular sieve, which is a LiX zeolite in the form of small beads (diameter = 0.5-0.8mm, strongly Geldart class B).It should be noted that Geldart class B particles have been selected rather than class A particles which provide a more uniform fluidization.This is done because of the higher heat and mass transfer when fluidizing class B particles, which suits these particles better in TCES (Rhodes 2008).Magnesium chloride salt, produced by Acros Organics, was purchased from Fischer Scientific.
In order to prepare five samples (250 g each) of the composite sorbent for testing, a similar methodology to that found in Xu et al. (2019) was used.1800 ml of deionized water was prepared, into which 200 g of Magnesium chloride was dissolved.This solution of 10 wt.% concentration was prepared by dissolving the anhydrous salt (MgCl 2 ) into the water using a hot plate and stir bar.The process of dissolving the salt into the water must be done a few grams at a time.This is in order to avoid both evaporation of the water due to heat released by the dissolving anhydrous salt and also to avoid large crystals from forming (the solution is also maintained at 80°C to ensure salt is dissolved).These crystals do not dissolve easily and will grow if not given ample time to dissolve.Once all the salt has been dissolved into the solution, the 1.25 kg of LiX zeolite particles are then immersed into the solution and left for 72 h to soak.The particles are then removed from the soak and rinsed from any excess solution.The particles are then placed in a thin layer on a tray in an electric, direct fired furnace to heat the particles evenly and activate them.The particles are brought to 200°C, then maintained at this temperature for 2 h.They are then allowed to return to ambient conditions in order to desorb them and ready them for "dry" fluidization testing.This process can be seen in Figure 3.The temperature should be below 208.5°C to avoid the unwanted production of Magnesium oxychloride and Hydrochloric acid (shown in Error!Reference source not found.),which has been calculated to begin at this temperature (at a vapor pressure of 2 kPa) (Xu et al. 2019).To check successful impregnation, EDX analysis is conducted to ensure values match those achieved by Xu et al. (2019).
A fluidized bed was designed and fabricated to test the particles under fluidization.The design was based on the general dimensions of a previously built fluidized bed at the Technical University of Munich (TUM) for the continuous reaction of TCES materials (Wuerth et al. 2019).However, as this bed is not designed for high-temperature reaction of >700°C as seen in the bed built at TUM, liberties could be taken in its construction.As such, the bed is constructed largely of rolled aluminum sheet for the body of the bed, and poly[methyl-methacrylate] for the bed walls.A test sieve with a mesh size of 53 µm manufactured by Retsch GmbH is located at the base of the bed to both distribute flow and prevent excess fines and particles from traveling below the bed.Above the test sieve is a perforated plate with an open area of 8.2% of the bed cross-sectional area, which is designed to fully distribute the flow and produce fluidization.This low open area has been selected as it has been shown that fluidization is more stable in a bed with a low perforated distributor due to pressure drop (Shukrie, Anuar, and Oumer 2016).After passing through the bed, air is vented through aluminum ducting and passed through several layers of G4 Grade filter media before venting to the atmosphere.An image and diagram of the bed (as well as the overall method, to be discussed subsequently) can be seen in Figure 4.
In order to take an accurate measure of mass flow rate through the bed, a combination of a mass flow controller (MKS GE50A) and a rotameter were used."18X" and "24" size rotameters were used to cover the entire range of flowrates measured (The 18 and 24 correspond to size of the rotameter opening, i.e., 18 mm and 24 mm.The "X" refers to the shape of the rotameter and is specific to the manufacturer (KDG)).This gave the ability to measure flow rates up to 215,000 SCCM (square cubic centimeters per minute) using the size 24 rotameter.The midrange was measured with the size 18X rotameter, and flow rates of up to 50,000 SCCM were measured with the mass flow controller (diagrammatically demonstrated in Figure 5).Above a maximum measurable flow rate, the bed begins to change from a fluidized bed to a pneumatic transport regime; this is highly undesirable and results in a significant loss of material.For this reason, a mechanical flow reducer attached to the rotameter will reduce the flow of gas if the rotameter maximum flow is reached.To ensure quality measurements, a highly accurate mass flow controller (±1 SCCM) was used to calibrate the rotameters in situ.
The fluidization of 250 g of the LiX zeolite is demonstrated in Figure 6.Here, it can be observed that incipient fluidization occurs at a superficial gas velocity of approximately 24 cm/s.Due to the Geldart class of the particles (Strongly class B particles), bubbling fluidization was visually observed almost immediately after incipient fluidization.This bubbling action can be seen in the pressure drop measurements, where large fluctuations in the measured pressure drop led to a wide standard deviation of the data points at higher superficial gas velocities.Despite this, the average pressure drop at each measured superficial gas velocity is still in keeping with the expected behavior of a fluidized bed.Pressure measurements were taken at a frequency of approximately 14 Hz, with the bed being held at each measured flow rate for 10-20 s.As the superficial gas velocity is decreased, the behavior of the defluidiation can be observed.As the bed is well aerated and well mixed after fluidization, the reduction in superficial gas velocity results in a reduction in pressure drop as the strength of the cohesive van der Waals force is reduced.However, as the superficial gas velocity reduced further agglomerated areas will begin to form, hence the quality of the fluidization (i.e., the amount of heat and mass transfer, as well as mixing) will be reduced (Singh 2016).
It can be seen in Figure 6 that while there is a clear regime change at close to the theoretical minimum fluidization velocity, the pressure drop indicated represents a low fluidization index of approximately 0.5.This likely results from two primary factors: the first is simply the placement of the pressure sensor at the base of the bed.It was not possible to locate the pressure sensor at the very base of the bed and so the inlet of this pressure sensor is located at approximately 20 mm above the distributor plate.This means that the measured pressure at the base of the bed will be slightly lower than the actual pressure at the base.The second factor which reduces the fluidization index is the strong bubbling nature of the bed, which has been observed to reduce the fluidization index (Sahoo and Sahoo 2013).Finally, the influence of some light spotting and channeling will reduce the pressure drop (it can be reduced by as much as half as a result of channeling when fluidization is poor) (Kunii and Levenspiel 1969).In order to verify these experimental data, the incipient fluidization velocity can be calculated using correlations produced by Baeyens and Geldart (Baeyens and Geldart 1974).This correlation is shown in Eq. 2, where ρ P , ρ f , g, d P , and μ are the densities of the particle and the fluid, gravity, particle diameter, and viscosity, respectively.Using this correlation, and assuming a particle density of approximately 1900 kg/m 3 ,  and a fluid density of 1.3 kg/m 3 , an incipient fluidization velocity of 27.31 cm/s can be calculated.This is in good agreement with the experimental data, which demonstrates an incipient fluidization velocity of 24.10 cm/s.These values can both be seen highlighted in Figure 6.(2020), a full cycle time of 40 min has been selected, with 20 min of "dehydration" and 20 min of "hydration" (although it should be noted that these steps are identical given that no hydration or dehydration is intended in this experiment).For each sample, only one cycle will be tested in order to isolate the variable of superficial gas flow rate alone.At the high gas velocity measured, a 40-min cycle would likely represent the amount of fluidization experienced by a batch for several cycles, rather than a single cycle.This is due to the small batch size and fast reaction time in a high velocity fluidized bed (Bardy et al. 2020).However, the quality of the energy produced by a TCES fluidized bed at high gas flow rates will of course be reduced.Both the pure LiX Zeolite, and the Composite Zeolite will be tested at four different gas flow rates, with a fifth sample from each not experiencing any flow at all (0 cm/s) to act as a control.The first sample is subjected to a superficial gas velocity of 15 cm/s which is not enough to produce fluidization, and as such investigates the particles' under-packed bed conditions.The next three samples are subjected to 30, 45, and 60 cm/s, respectively.This represents the range of fluidization velocities from just after incipient fluidization, a middle range, and just before pneumatic transport begins.This will demonstrate whether, and at what point in the range of superficial gas velocities, any degradation occurs.Hence, 10 samples are tested; 5 pure zeolite and 5 composite.After testing, these samples were placed in airtight containers and sealed with tape to prevent any fines from escaping.Once all tests had been completed, the samples were then observed under a Zeiss Stemi 508 Greenough Stereo Microscope for the observation of the particles.This method for measuring particle size distribution has been validated for various particles (Goel et al. 2018).It has also been validated for use in Figure 6.Superficial gas velocity against pressure drop in an 74mm diameter, 140mm high-fluidized bed of 0.5-0.8mmdiameter LiX Zeolite (250g), (1 standard deviation shown).
analyzing the particle size distribution of nanoparticles (Wierzbinski et al. 2018).As per these studies, a few hundred particles have been measured for each sample.To ensure that each sample is mixed, the sample is place in a container and rotated gently (to avoid incurring degradation through this rotation), before a few thousand particles are placed on a dish for measurement.For each sample four images are captured at a consistent magnification in different regions of the sample, and a scale digitally added to the image.In order to observe the fines, the samples are observed under a Zeiss Stemi 8-40× 305 stereo microscopes.These images are then processed one at a time through the open architecture "ImageJ" image processing program.The scale of the image is then measured five times in the software and an average taken, this is then used to calculate the ratio of pixels to microns (or nanometers when observing the fines) to set the scale of the image for the software.A threshold process is then conducted on the image which divides the image into two classes of pixels, representing the background (i.e., pixels which are not of the particles), and foreground (i.e., pixels which are the particles).The software analyzes the image to calculate the area of each distinct cluster of pixels.The software is set to measure only clusters with a circularity of 0.6 so as not to measure a row of adjoined clusters of pixels, or the pixels representing the scale bar; this can be seen in Figure 7.It should be noted that this method of measurement is highly accurate, with an average error between the software and a graticule (a small piece of glass with a highly accurate measurement scale on its surface) of 0.1% being found.The error arising from resolution of the measurement is 3.4 µm.The value measured for each of these clusters of pixels is the area of the cluster, with each image providing approximately 100 measured particles.Assuming the cross section of each particle is approximately circular, the diameter of each particle is then calculated using the area values.A similar method is conducted on the images of the fines, however due to the irregular shape of these, a circularity threshold of 0.3 is used.As these particles tend not to conglomerate, the issue of measuring multiple particles as one large particle does not present itself.This overall methodology can be seen in Figure 4.

Results and discussion
EDX analysis conducted on the zeolites prior to and post impregnation show successful impregnation.The ratio of magnesium to aluminum increased from 0 g/g to 0.415 g/g.This corresponds to the values achieved by Xu et al. (2019), who found ratios of 0.388-0.525g/g for solution concentrations between 9% and 13%.The impregnation process has been observed to increase the average particle diameter by 7%−8%, from a pre-impregnation diameter of 0.6354 (±0.0529, σ = 1) mm to a post-impregnation diameter of 0.6835 (±0.0522, σ = 1) mm, these values can be seen in Figure 8.While the standard deviation for these measurements is significant, both measurements' average falls outside one standard deviation from the other.This increase in diameter is attributed in part to the particles not being fully desorbed during the baking process, which is also evidenced by some particles remaining the dark color which they take on after adsorption even after baking.This can be seen in Figure 9, where the discolored particles are circled in red.It should be noted that the color of the other non-circled particles differs in the two images due to a lighting error and is visually the same color.As such, these discolored particles likely have water sorbed into their bulk and have increased in diameter for this reason, as observed by Xu et al. (2019).Now that the average diameter of the particle has been established, this data can be used as a foundation to assess the degradation of the particles after fluidization.The finding that the impregnation process increases particle diameter is in agreement with the suggestion by Xu et al. (2019) that the zeolite matrix expands following hydration.
Another reason for the increase in average diameter after impregnation is the formation of some large salt crystals on the surface of the particles.These salt crystals have also led to cohesion of small clusters of particles, although care has been taken not to allow these clusters to affect the diameter measurement.Both of these phenomena can be seen in Figure 10.The methodology should in future be adapted to remove this excess salt through further rinsing.
These salt particles will increase the measured average diameter of the particles.They will also change the fluidization properties of the particles.The salt crystals could cause the particles to be more cohesive and as such closer to Geldart class C particles, which as previously discussed would be unfavorable in a TCES system.As such, it is important to observe whether there is a significant effect on the fluidization characteristics of the composite particles compared to the standard zeolites, and also whether the salt crystals are present after fluidization.
It has been shown that the impregnation process affects the particle diameter.However, the fluidization of the particles for a 40-min cycle (comprised of two 20-min periods of gas flow), has no measured effect on the particle diameter (Figure 11).This is true for LiX Zeolite with no impregnation, and for LiX Zeolite impregnated with magnesium chloride.This is also true across the range of superficial gas velocities tested.It can also be said that the flow of gas over these particles at sub-fluidization velocities also does not significantly affect the particle diameter.For both the LiX Zeolite with no impregnation and for LiX Zeolite impregnated with magnesium chloride, the trend of particle size after fluidization at each gas velocity is approximately flat and within the deviation indicating no change in average particle size.
The values of the error bars are given in Figure 11 Average diameter of LiX zeolite compared against a Composite LiX magnesium chloride zeolite after one 40-min cycle, against superficial gas velocity (each data point indicates a different sample, cycled at a different superficial gas velocity) (error bars indicate one standard deviation) Table 1, these values represent one standard deviation and as such are calculated using Eq. 3.Where "σ," "N," "x i ," and "µ" represent the standard deviation, the size of the population of measurements, the variable value of each measurement, and the mean value, respectively.As is discussed previously, the error arising from the measurement is negligibly small.σ ¼ ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi P This data shows that these composite particles are unlikely to suffer fragmentation through the stresses produced during fluidization at any gas velocity (extremely high gas velocities resulting in pneumatic transport are not investigated).The particles are also not experiencing enough abrasion to significantly reduce their diameter.Again, this is true for both the composite and non-composite particles.However, this is not to say that no fines were produced; initial observations showed that the prevalence of fines visually observed after fluidizing LiX zeolite for a 40-min cycle increased proportionally to the superficial gas velocity.However, production of fines was also produced at superficial gas velocities below that of the minimum fluidization velocity (i.e., In a fixed bed).This observation would suggest that the production of fines cannot be entirely attributed to the action of fluidization.Another observation which lends credit to this is that the action of simply pouring the particles into the bed prior to any testing resulted in the production of some fines also.This is also true for the composite particles which also appeared to produce fines at all ranges of superficial gas velocities, but also in the transport of the particles in and out of the bed for testing.For both composite and non-composite particles, increasing the gas velocity appeared to increase the rate of fine production.This is to be expected as a result of both an increase in the stresses experienced by the particles as higher velocities result in higher force impacts with the wall and with other particles, but also because of the increase in the number of impacts experienced.This is in agreement with prior research investigating attrition in fluidized beds where the rate of attrition is directly proportional to the superficial gas velocity (Zhang et al. 2016).
The distribution of the diameter of the fines produced by LiX Zeolite of mesh 18-35 after fluidizing at a superficial gas velocity of 60 cm/s is shown in Figure 12.Fines in the smallest diameter bin are between 530 nm and 1140 nm in diameter, and thus represent the limit of fines that can be observed through optical microscopy (due to the wavelength of light).The mean diameter of these fines was 3.60 μm (σ = 1, ±2.34 μm).While these fines will eventually erode the particles and reduce their diameter, it is at a rate not detectable with this methodology, and as such is unlikely to be relevant over several cycles at the lower gas velocities.Also, no loss of mass was detected after a 40-min cycle at any of the superficial velocities tested for all batches of composite and non-composite particles.However, due to the resolution of the scale a mass loss lower than 0.125% of the total mass would not be detectable.As such, if a maximum mass loss is assumed, the mass loss resulting from fluidization at the highest superficial gas velocity would represent 1% of the total mass after eight cycles.A final point to note with regard to the fines produced, is that significant care should be taken to properly filter these fines, as research has shown that particles <1 µm travel deepest into the lungs and can pose a significant health hazard (EEA 2022).
Another minor observation during testing was cohesion of particles to the wall (poly[methylmethacrylate]).A small proportion of particles would be thrown from the surface of the bed during bubbling fluidization and attached to the wall.This is likely due to the generation of static forces during fluidization, which has first been observed to occur during fluidization by Miller and Logwinuk (Miller and Logwinuk 1951), then studied at length by Boland and Geldart (Boland and Geldart 1972).As stated by Wolny and Opalinski (Wolny and Opalinski 1984), this static electricity generation can have an effect on the hydrodynamics of the bed and thus affect the heat transfer of immersed heat exchange surfaces.It has been shown that deionizing or humidifying the fluidizing gas can eliminate this buildup of static charge (Revel et al. 2003).If the discharge of the TCES materials studied here involves the introduction of humid air, then the issue of static charge buildup is unlikely to affect the heat transfer within the bed during use.It is important to note that in no samples has any cracking been observed, which shows that the cracking and fragmentation of the zeolite observed in prior research is likely originating purely from the hydration and dehydration of the particles increase and decreasing the particle diameter and surface area between cycles.A process that will incur significant surface and internal stress on the particles.
One final interesting observation when examining samples of the composite particles under the microscope was the presence of large salt crystal clusters up to 200 µm in diameter.These clusters of crystals are similar to the crystal which can be seen adhered to the particle in Figure 10, however these crystals can be seen unattached to any particle.Examples of this can be seen with large salt crystals and regions with several salt crystals circled in red in Figure 13.

Conclusion
In this study, LiX Zeolite has been impregnated with MgCl 2 and the effect of fluidization at different gas velocities of the LiX Zeolite and of a composite MgCl 2 -LiX Zeolite have been investigated.It has been shown that:    • The impregnation methodology increases the diameter of the zeolite particle from a preimpregnation diameter of 0.6354 (±0.0529, σ = 1) mm to a post-impregnation diameter of 0.6835 (±0.0522, σ = 1) mm (approximately 7.57%).• Fluidization at superficial gas velocities up to 60 cm/s does not generate the necessary internal stress to fragment both LiX zeolites, and composite zeolites.
• Both composite and non-composite zeolites appeared to experience abrasion with fines being observed in the filter media after fluidization in both packed and fluidized tests.As such, it cannot be said that fluidization alone is responsible for this abrasion.
• No reduction in average diameter at all gas velocities occurred as a result of this abrasion.In addition, neither set of particles experienced a detectable loss of mass after fluidization.• Fines produced, were analyzed and found to have a diameter of 3.60 μm (σ = 1, ±2.34 μm).
• Fines produced are of a size where a risk is posed if inhaled.
Future research should be conducted to ascertain the influence of the charge and discharge phases of TCES on its performance in a fluidized bed and vice versa.The addition of moisture into the fluidizing gas could have several potential effects.The gas will change in average density and viscosity, which may influence fluidization.However, a greater effect of fluidization is likely to be brought about by the cohesive forces present when moisture is added, with phenomena such as caking and conglomeration potentially adversely affecting the fluidized bed reactor.It will also be important to study whether in the long term (i.e., over many cycles) the small amount of abrasion observed incurs a significant reduction in particle diameter and as such its suitability for fluidization.If the particles experience strong degradation through abrasion over time, then they will be unlikely to be suited to fluidization as the gas flow will entrain the fines produced and reduce the bed mass, presenting a health hazard.Study of the micro degradation of the particle surface using SEM imagery would also be beneficial in assessing whether macro-degradation is likely to be brought about from micro-and meso-cracks.Finally, while the findings presented here are likely to be applicable to other composite TCES materials.Due to the lack of other studies, further study must be conducted to verify this, this would be valuable so as to avoid the use of lithium which is becoming increasingly scarce.

Figure 1 .
Figure 1.Diagrammatic demonstration of the primary mechanisms of particle degradation.

Figure 2 .
Figure 2. Images showing slugging and channelling of Geldart class C particles in a fluidised bed.

Figure 3 .
Figure 3. Schematic illustration of the composite TCES production process.

Figure 4 .
Figure 4.An image (a) and diagram (b) of the fluidised bed reactor, and a schematic presentation (c) of the overall methodology.
. (2017) tested a TCES, CaO/Ca(OH) 2 , reaction in a fluidized bed and ran the bed for approximately 30 min during hydration and 30 min during dehydration.Pardo et al. (2014) ran a fluidized bed for a similar time in a similar setting.Bardy et al. (2020) fluidized zeolite 13X for approximately 50 min at a rate of 30 L/min.Given that flow rates exceed those tested by Bardy et al.

Figure 7 .
Figure 7. Image of Zeolite particle prior to analysis in ImageJ (Left); image after analysis in ImageJ, with measured particles shown with a cyan outline (right).

Figure 8 .
Figure 8. Diameter distribution of particles before (top) and after (bottom) impregnation.Bell curves are shown in orange for the pure LiX zeolite, and green for the composite.

Figure 9 .
Figure 9. Images of LiX Zeolite post impregnation and pre impregnation with MgCl 2 with discolored particles highlighted in with red circles.

Figure 10 .
Figure 10.Salt crystals forming on the surface of zeolite particles after impregnation process.

Figure 11 .
Figure 11.Average diameter of LiX zeolite compared against a Composite LiX magnesium chloride zeolite after one 40-minute cycle, against superficial gas velocity (each data point indicates a different sample, cycled at a different superficial gas velocity) (error bars indicate one standard deviation).

Figure 12 .
Figure 12.Distribution of fines produced by LiX Zeolite of mesh 18-35 after fluidising at a superficial gas velocity of 60 cm/s.