Experimental Characterization of Phase Change Materials for Refrigeration Processes

: Latent heat storage units for refrigeration processes are promising as alternatives to water/glycol-based storage due to their signiﬁcantly higher energy densities, which would lead to more compact and potentially more cost-effective storages. In this study, important thermophysical properties of ﬁve phase change material (PCM) candidates are determined in the temperature range between − 22 and − 35 ◦ C and their compatibility with relevant metals and polymers is investigated. The goal is to complement existing scattered information in literature and to apply a consistent testing methodology to all PCMs, to enable a more reliable comparison between them. More speciﬁcally, the enthalpy of fusion, melting point, density, compatibility with aluminum, copper, polyethylene (PE), polypropylene (PP), neoprene and butyl rubber, are experimentally determined for 1-heptanol, n-decane, propionic acid, NaCl/water mixtures, and Al(NO 3 ) 3 /water mixtures. The results of the investigations reveal individual strengths and weaknesses of the ﬁve candidates. Further, 23.3 wt.% NaCl in water stands out for its very high volumetric energy density and n-decane follows with a lower energy density but better compatibility with surrounding materials and supercooling performance. The importance of using consistent methodologies to determine thermophysical properties when the goal is to compare PCM performance is highlighted.


Introduction
Approximately 48 Terawatt-hours of the European Union final energy demand are dedicated to process cooling at temperatures below 0 • C [1]. The vast majority of this cooling demand is covered by means of compression cooling using electricity as the energy carrier [2]. This includes the chemical, pharmaceutical, food, and transportation sectors. Thermal storage has been identified as a technology that could decrease the installation cost of refrigeration and cooling systems [3], increase the efficiency and operation cost of refrigeration applications [4], and support the integration of renewables in refrigeration processes [5]. The state of the art, in commercial refrigeration processes, are sensible storage solutions based on water/glycol mixtures [6]. This technology, even though well established, has various disadvantages, such as: low energy density and therefore high storage volumes, high cost, and potential limitations in power input and output if mass flow rate restrictions need to be applied to increase stratification efficiency [7].
Latent heat storage has been proposed as an alternative storage method for refrigeration processes in some studies, mostly in the context of: (i) incorporation of phase change material (PCM) in freezers, (ii) transport of food and temperature-sensitive goods, and (iii) substitution of water/glycol storage in industrial refrigeration processes [8]. In most of these studies, mixtures of NaCl and water were proposed [9] because of their good thermophysical properties and low cost. Different types of salt/water mixtures and paraffins have also been proposed as PCMs for refrigeration processes [9][10][11]. Recently, Suamir et al. [12] proposed corn-oil/water mixtures as PCMs for refrigeration processes with melting temperatures and enthalpies ranging from 0 to −27 • C and 297.4-68.7 J/g for concentrations of corn-oil ester in water ranging from 0-35%, respectively. Several commercial PCMs of unknown compositions also exist in the market at temperatures suitable for refrigeration processes [13]. For most of the investigated PCMs, melting point, enthalpy of fusion, and density have been reported in literature [11,14,15]. However, the values reported are scattered and measured under different conditions and are in many cases incomplete (for example degree of supercooling is rarely reported). At the same time, almost no concise information is presented in literature with respect to the compatibility of these PCMs with metals and plastics even though this information is essential for their implementation in latent heat storage applications.
The goal of this study is to determine the important thermophysical properties of five PCM candidates in the temperature range between −22 and −35 • C and present their compatibility with relevant metals and plastics. This way, existing scattered information in literature will be complemented and the consistent methodology applied for all PCMs will enable a more reliable comparison between them. In this work, we investigate 1-heptanol, ndecane, propionic acid, NaCl/water mixtures, and Al(NO 3 ) 3 /water mixtures. The enthalpy of fusion, melting point, density, compatibility with aluminum, copper, polyethylene (PE), polypropylene (PP), neoprene and butyl rubber, are experimentally determined and presented in this paper. The research work was performed in the framework of the Eurostars project STOREF with the ultimate goal of developing a novel, high power latent heat storage for refrigeration applications. Table 1 shows the information of the raw materials used as a basis for PCMs in this study. The organic materials were used for analysis directly without processing and purification. Sodium chloride and aluminum nitrate nonahydrate were mixed with the appropriate amount of deionized water to achieve the final concentration of NaCl 23.3 wt.% and Al(NO 3 ) 3 30.5 wt.%, respectively. The aluminum, copper, polyethylene (PE), polypropylene (PP), neoprene and butyl rubber used for the material compatibility tests were provided by Sunamp Ltd. (East Lothian, UK). and are the same materials that are used in their heat batteries. The PE used is a high temperature grade and the PP is a low-density linear grade.

Determination of Phase Change Temperature and Phase Change Enthalpy
Differential scanning calorimetry (DSC) measurements were performed using a Mettler Toledo DSC823 device with auto-sampler to determine the phase change temperatures and the phase change enthalpies of the five PCM candidates. The device was calibrated and checked prior to the measurements using indium and zinc, following the specifications provided by the device manufacturer. Heating rates (β) of 10 and 2 K/min were chosen for the determination of the phase change enthalpy and the phase change temperature, respectively (see Figure 1).

Determination of Phase Change Temperature and Phase Change Enthalpy
Differential scanning calorimetry (DSC) measurements were performed using a Mettler Toledo DSC823 device with auto-sampler to determine the phase change temperatures and the phase change enthalpies of the five PCM candidates. The device was calibrated and checked prior to the measurements using indium and zinc, following the specifications provided by the device manufacturer. Heating rates (β) of 10 and 2 K/min were chosen for the determination of the phase change enthalpy and the phase change temperature, respectively (see Figure 1).

Figure 1.
Example of DSC melting curves resulting from measurements with (a) 10 K/min to determine the phase change enthalpy and (b) 2 K/min to determine phase change temperature. In both cases, all three repetitions are plotted in the same graph, showing the very consistent melting behavior that n-decane presented.
The lowest temperature was set at −60 °C (which also coincided with the lowest limit of the device) and the highest temperature at 10 °C to guarantee that the entire samples were melted prior to the next cooling cycle. The experiments were conducted using hermetically sealed, 40 μL aluminum crucibles under nitrogen atmosphere with a gas flow Figure 1. Example of DSC melting curves resulting from measurements with (a) 10 K/min to determine the phase change enthalpy and (b) 2 K/min to determine phase change temperature. In both cases, all three repetitions are plotted in the same graph, showing the very consistent melting behavior that n-decane presented.
The lowest temperature was set at −60 • C (which also coincided with the lowest limit of the device) and the highest temperature at 10 • C to guarantee that the entire samples were melted prior to the next cooling cycle. The experiments were conducted using hermetically sealed, 40 µL aluminum crucibles under nitrogen atmosphere with a gas flow of 50 mL/min. Each sample underwent three heating and cooling cycles at 2 K/min and three heating and cooling cycles at 10 K/min. The melting (T m ) and crystallization (T c ) temperatures were calculated by the STARe software through the tangent method, and the enthalpies (∆H) were obtained by integration of the corresponding peaks. The degree of supercooling, ∆T, was calculated using the difference between the onset melting temperature and the onset crystallization temperature reported by the instrument. T m , T c were determined using the cycles conducted at 2 K/min while ∆H was determined based on the cycles conducted at 10 K/min. The method used is based on the one developed in [16]. The methodology for the enthalpy determination was adapted slightly for the Al(NO 3 ) 3 /water mixtures as the solidification temperature was very close to the lower temperature limits of the DSC device and the dynamic cooling alone was at times not adequate to solidify the entire sample. Therefore, an additional isothermal step at −60 • C was applied for 10 min at the end of the dynamic cooling phase to ensure complete solidification of the sample.

Determination of Density
The densities were determined at room temperature (25.9 • C) while all samples were in a liquid state. Initially, 1 mL of the PCM was measured using a 1 mL GILSON pipette. The sample was subsequently placed in a weighing balance (Mettler Toledo, Columbus, OH, USA, Model MT5, series number L59982) where its mass was measured. The density was calculated by dividing the sample mass by the sample volume. The measurements were repeated three times. The values reported are the mean values obtained from the repetitions.

Determination of Material Compatibility
All PCM candidates were tested with respect to their compatibility with typical materials they could come in contact within a latent heat storage application. In particular, the PCMs were tested in combination with potential metallic heat exchanger materials (aluminum, copper, and their combination) and surrounding plastic and rubber samples (polypropylene (PP), polyethylene (PE), neoprene and butyl rubber).
A limited number of investigations can be found in literature regarding the compatibility of PCMs with various metals and plastics in general [17][18][19] and even less so for PCMs used in cooling applications in particular [8]. There is no standardized methodology for material compatibility tests in the latent heat storage community and most researchers develop their own testing procedure. In this study, it was decided to follow standards that have been developed and established by the American Society for Testing and Materials (ASTM G31 and ASTM D543). Figure 2 shows an example of the untreated metallic samples. To determine the corrosion rate of the different metals in contact with the chosen PCMs, immersion tests of the metallic samples in the PCMs were performed. For each PCM, three tests were conducted. One with aluminum, one with copper, and one with a combined aluminum/copper sample to account for galvanic effects. The cylindrical copper samples had a fixed wall thickness (H) and a fixed outer diameter (W). The aluminum samples were cut into rectangular shapes with a width (W) of 15 mm and a wall thickness (H) of 0.1 mm. The surface of the extra aluminum surrounding the hole in the center of the samples was considered to be equal to the material that would be needed to cover the hole. Therefore, the aluminum samples were handled as perfect rectangles. The third dimension of the samples designated here as length (Li) depended on the sample cutting and was calculated based on the sample mass and metal density as follows: where Vi was the sample volume, mi was the measured sample mass, and ρ was the den- The third dimension of the samples designated here as length (L i ) depended on the sample cutting and was calculated based on the sample mass and metal density as follows: where V i was the sample volume, m i was the measured sample mass, and ρ was the density of the corresponding metal (8.96 g/cm 3 for copper and 2.7 g/cm 3 for aluminum).
To calculate the initial surface area of the samples, S i , the following equation was used for both metals: All metallic samples were cleaned according to ISO 8407:2009 (for about 2-5 min). Subsequently, they were weighed in a balance (Ohaus EP214C) and then placed in a 50 mL falcon tube containing 20-30 mL PCM. The size of the specimens and the amount of PCM were chosen according to standard ASTM G31 [20], which recommends a solution volume to specimen area ratio of 0.20 mL/mm 2 for single metallic samples. The immersion test was carried out at room temperature for 42 days. After the immersion tests were completed, the metallic samples were extracted from the PCM, they were mechanically cleaned with a soft toothbrush, to remove any corrosion products in order to measure the weight of the remaining pure metal. After drying, the weight of the samples was recorded. Subsequently, the samples were subjected to chemical cleaning in order to remove any remaining corrosion products, based on standard procedures [21]. For the chemical cleaning, aluminum samples were placed in nitric acid 70 wt.% and the copper samples were placed in HCl 23 wt.%. The samples were left in their corresponding acid for 80 s in total and then removed, dried, and weighed at regular intervals of 10 to 20 s to distinguish between the removal of the corrosion product as opposed to pure metal, as explained in [21]. The average, area-specific corrosion rate was then calculated as follows: where m e is the final sample mass after removing all corrosion product and t is the time of exposure of the metal to the PCM. For the different heat and cold batteries, Sunamp utilizes polypropylene, polyethylene, or butyl rubber as encapsulation material and neoprene as a sleeve (see Figure 3).  For the purpose of this compatibility study, samples of about 1 cm 2 were taken from each of the four types of plastic/rubber. The specimens were weighed with a balance (Ohaus EP214C) and their initial mass, mi, was recorded. The sample was then placed into a 20 mL glass tube and filled with 10 mL PCM. The guideline ASTM D543 recommend that "the quantity of reagent shall be approximately 40 mL/inch 2 of specimen surface area" [22]. Each sample was tested in duplicate. Samples were stored at room temperature for one week, after which the samples were extracted, rinsed with deionized water, dried, and immediately reweighed and the weight, me, was recorded. The samples where then left to dry at room temperature for an additional week and the mass, me,d, was recorded. Based on these measurements, the mass loss or gain after drying was calculated. The loss For the purpose of this compatibility study, samples of about 1 cm 2 were taken from each of the four types of plastic/rubber. The specimens were weighed with a balance (Ohaus EP214C) and their initial mass, m i , was recorded. The sample was then placed into a 20 mL glass tube and filled with 10 mL PCM. The guideline ASTM D543 recommend that "the quantity of reagent shall be approximately 40 mL/inch 2 of specimen surface area" [22]. Each sample was tested in duplicate. Samples were stored at room temperature for one week, after which the samples were extracted, rinsed with deionized water, dried, and immediately reweighed and the weight, m e , was recorded. The samples where then left to dry at room temperature for an additional week and the mass, m e , d , was recorded. Based on these measurements, the mass loss or gain after drying was calculated. The loss (or gain) of mass was calculated as follows: The mass change could be positive or negative depending on whether the sample gained or lost weight during the experiments. Table 2 presents the densities of the five PCMs as they were measured in the framework of this study and reported in literature. The experimental results show a very good precision and a very good agreement with the values presented in literature. Therefore, these values can be trusted and utilized for the calculation of the volumetric enthalpies of the PCM. As expected, salt/water mixtures present higher densities than organic substances. Propionic acid also has a rather high density, a bit lower than 1 kg/L.  Table 3 summarizes the phase change temperatures and enthalpies as they were determined in the framework on this study and reported in literature. It is worth mentioning that only melting (no solidification) temperatures could be found in literature. In most cases, it was also not mentioned if they referred to onset or peak melting points. In this investigation, it was considered useful to report both onset and peak temperatures for both the melting and crystallization processes as an indication of the ability of the PCM to undergo phase change in a relatively narrow temperature range. Additionally, the supercooling degree in the DSC is reported as an indication of the tendency of each PCM to supercool, keeping in mind that the supercooling in the actual application will likely differ from the one reported here, as supercooling has been observed to be a volume-dependent phenomenon [26]. Additionally, both the volumetric and the gravimetric phase change enthalpies are reported as both could be crucial, especially when we consider stationary vs. mobile applications. The PCMs could be separated into two regions, based on their melting point. Propionic acid and NaCl/water would be suitable for higher temperature applications (>−25 • C) and the remaining PCMs for lower temperature applications (<−30 • C). All PCMs present a relatively narrow melting temperature range, which in all cases, remained below 5 • C. Most PCMs present a rather high supercooling tendency (>10 • C) with the exception of n-decane. The rather high standard deviations of both melting and solidification points of NaCl/water should be noted as they could indicate a tendency of the mixture to undergo different crystallization processes (e.g., formation of different crystal structures due to mixing and/or stochastic nucleation effects). This is supported by a closer inspection of the DSC diagram of the crystallization process of the NaCl/water mixture, which as can be seen in Figure 4, presented numerous peaks. The PCMs could be separated into two regions, based on their melting point. Propionic acid and NaCl/water would be suitable for higher temperature applications (>−25 °C) and the remaining PCMs for lower temperature applications (<−30 °C). All PCMs present a relatively narrow melting temperature range, which in all cases, remained below 5 °C. Most PCMs present a rather high supercooling tendency (>10 °C) with the exception of ndecane. The rather high standard deviations of both melting and solidification points of NaCl/water should be noted as they could indicate a tendency of the mixture to undergo different crystallization processes (e.g., formation of different crystal structures due to mixing and/or stochastic nucleation effects). This is supported by a closer inspection of the DSC diagram of the crystallization process of the NaCl/water mixture, which as can be seen in Figure 4, presented numerous peaks. Similar effects were observed in the case of the Al(NO3)3/water PCM in all three repetitions of the cooling curves with Al(NO3)3/water 30.5 wt.% as they were measured by DSC at 2 K/min (see Figure 5). The presence of multiple and shifted peaks during cooling also indicate the formation of various phases during crystallization that could be attributed to supercooling and mixing effects. It can also be observed that crystallization is still ongoing when the device reaches its lower temperature limit (−60 °C) which required the addition of an isothermal step at the end of the process to ensure complete sample solidification. Further studies in larger devices, where less supercooling is present and/or with the presence of nucleation or gelling agents, can be conducted to further investigate Similar effects were observed in the case of the Al(NO 3 ) 3 /water PCM in all three repetitions of the cooling curves with Al(NO 3 ) 3 /water 30.5 wt.% as they were measured by DSC at 2 K/min (see Figure 5). The presence of multiple and shifted peaks during cooling also indicate the formation of various phases during crystallization that could be attributed to supercooling and mixing effects. It can also be observed that crystallization is still ongoing when the device reaches its lower temperature limit (−60 • C) which required the addition of an isothermal step at the end of the process to ensure complete sample solidification. Further studies in larger devices, where less supercooling is present and/or with the presence of nucleation or gelling agents, can be conducted to further investigate these effects. It has to be noted that in contrast to the inorganic PCMs, all organic PCMs exhibited single phase transformations and very good repeatability.

Measurement of the Thermo-Physical Properties
While the measured melting points are in relatively good agreement with the values reported in literature, the same cannot be stated for the melting enthalpies, where 1heptanol and Al(NO 3 ) 3 /water deviated from the literature values by −25.7% and 32.5%, respectively. Such deviations are often observed in literature and can be attributed to utilization of different measurement methodologies (e.g., instrumentation, heating rates, inclusion or not of sensible heat, etc.).
Finally, NaCl/water presents the best performance in terms of gravimetric phase change enthalpy, followed by n-decane. The performance of the former is even more impressive in terms of volumetric phase change enthalpy where it presents a 33% higher enthalpy value with respect to the second best substance, Al(NO 3 ) 3 /water. these effects. It has to be noted that in contrast to the inorganic PCMs, all organic PCMs exhibited single phase transformations and very good repeatability. Finally, NaCl/water presents the best performance in terms of gravimetric phase change enthalpy, followed by n-decane. The performance of the former is even more impressive in terms of volumetric phase change enthalpy where it presents a 33% higher enthalpy value with respect to the second best substance, Al(NO3)3/water.

Metallic Samples
The main characteristics of the initial metallic samples, their absolute mass loss and mass loss % after mechanical and after mechanical and chemical cleaning, as well as the calculated corrosion rates are shown in Table 4. The most severe corrosion was observed for the Al(NO3)3/water samples in combination with Al both in the pure Al samples and in combination with Cu. In both cases, the Al sample was completely destroyed before the end of the immersion experiments and was no longer visible. Similar, slightly higher absolute mass losses were also observed for the Cu samples in Al(NO3)3/water samples, but since the Cu samples were heavier to begin with, a part of the samples was still present at the end of the tests.

Metallic Samples
The main characteristics of the initial metallic samples, their absolute mass loss and mass loss % after mechanical and after mechanical and chemical cleaning, as well as the calculated corrosion rates are shown in Table 4. The most severe corrosion was observed for the Al(NO 3 ) 3 /water samples in combination with Al both in the pure Al samples and in combination with Cu. In both cases, the Al sample was completely destroyed before the end of the immersion experiments and was no longer visible. Similar, slightly higher absolute mass losses were also observed for the Cu samples in Al(NO 3 ) 3 /water samples, but since the Cu samples were heavier to begin with, a part of the samples was still present at the end of the tests. * In these cases, the metallic samples had been completely destroyed before the end of the experiment. The corrosion rates would therefore be higher than the ones reported here.
The summary of the resulting corrosion rates can be seen in Figure 6. The suitability of the PCM/metal combinations can be then evaluated using the values reported in Table 5. The combined information shows that the Al(NO 3 ) 3 /water PCM is not suitable for implementation in a latent heat storage application where it would be in contact with any copper or aluminum parts. Propionic acid is also found to have a critically high corrosion rate with copper. As latent heat storage units should have a life-time of at least 10 years, propionic acid is regarded as an unsuitable PCM for latent TES where the PCM is in contact with copper. It should be noted that in the case of NaCl/water in contact with Al, optical observations revealed the occurrence of pitting corrosion. This was to a degree expected due the presence of dissolved Cl − in contact with aluminum metal, a well-known corrosion scenario. Therefore, aluminum is considered unsuitable to be used both in the storage tank and the heat exchanger with this PCM, despite the low corrosion rates. As the utilized standards assume a uniform attack over the entire metallic surface and cannot account for pitting corrosion, the calculated corrosion rates cannot be solely relied upon in the specific material combination. The remaining two PCM candidates presented an acceptable corrosion rate with both Al and Cu and are considered suitable for latent heat storage applications where these metals are present. * In these cases, the metallic samples had been completely destroyed before the end of the experiment. The corrosion ra would therefore be higher than the ones reported here.
The summary of the resulting corrosion rates can be seen in Figure 6. The sui of the PCM/metal combinations can be then evaluated using the values reported i 5. The combined information shows that the Al(NO3)3/water PCM is not suitable plementation in a latent heat storage application where it would be in contact w copper or aluminum parts. Propionic acid is also found to have a critically high co rate with copper. As latent heat storage units should have a life-time of at least 10 propionic acid is regarded as an unsuitable PCM for latent TES where the PCM is tact with copper. It should be noted that in the case of NaCl/water in contact w optical observations revealed the occurrence of pitting corrosion. This was to a expected due the presence of dissolved Cl − in contact with aluminum metal, a wellcorrosion scenario. Therefore, aluminum is considered unsuitable to be used both storage tank and the heat exchanger with this PCM, despite the low corrosion ra the utilized standards assume a uniform attack over the entire metallic surface and account for pitting corrosion, the calculated corrosion rates cannot be solely relie in the specific material combination. The remaining two PCM candidates presen acceptable corrosion rate with both Al and Cu and are considered suitable for late storage applications where these metals are present.    [30]).

>1000
Completely destroyed within days 100-999 Not recommended for service greater than one month 50-99 Not recommended for service greater than one year 10-49 Caution recommended <9. 9 Recommended for long term service It is worth mentioning that the methodology utilized for the corrosion tests enabled the determination of only one average corrosion rate, which accounted for the events occurring during the entire immersion tests. However, in reality, the corrosion rate typically exhibits a transient behavior, and presents a higher value during the first days of immersion, which is reduced over time and reaches a steady, lower corrosion rate after some days/weeks. The authors chose to use this average corrosion rate as an indicator of the material compatibility as it represents the maximum possible value for this specific temperature level (25 • C). This is therefore a conservative, worst-case scenario approach, which gives confidence that the materials determined as compatible will perform even better than expected. This statement is also supported by the fact that the tests were conducted at room temperature even though the storage operation will be performed at temperatures well below 0 • C. The experiments were designed to account for a worst-case scenario, where the storage might not be in operation for longer periods of time and therefore exposed to ambient temperatures.

Polymeric Samples
The results of the plastic samples immersed in the five PCMs are summarized in Figure 7 and are shown in more detail in Table 6. Figure 7 (left) shows the mass gain of each plastic sample directly after its extraction, rinsing and drying (day 0 after conclusion of the immersion test), which is directly correlated to the degree of uptake of the PCM by the polymer. Whereas small degrees of swelling can be acceptable, higher degrees can cause alteration of the thermomechanical properties of the polymer, pressure increase and also leakage of the PCM through the polymeric enclosure. Figure 7 (right) shows the sample mass gain/loss with respect to its initial mass after 7 days of drying. These results should indicate whether the swelling was reversible, and also whether extraction of soluble constituents took place. On day 0 of drying, neoprene showed a very high mass uptake of the organic PCMs (118-431%) and a significant uptake of the inorganic ones (15-24%). It is therefore considered unsuitable for use in storage units especially with organic PCMs, as it would experience very high swelling. Butyl rubber showed a significant uptake of n-decane and propionic acid (38-85.5%) and it is therefore considered unsuitable for utilization with these PCMs. n-Decane also showed high migration in polypropylene, corresponding to a 34% mass gain of the PP sample, and therefore this combination of this polymer and PCM is also not advisable. The rest of the polymer/PCM combinations exhibited a low mass variation and are considered promising combinations for storage applications. Figure 7 (right) presents the mass variations of the same samples, 7 days after they were extracted from the immersion tests. Interestingly, the butyl rubber samples showed significant mass loss with respect to the original value (>4%) for the immersion test with all organic PCMs. This indicates removal of soluble constituents from the butyl rubber which could be observed at day 7, after the uptake of PCM in the polymeric samples has been reversed. This mass loss is considered significant and should be further investigated to determine its effects on the durability of the polymer. This effect highlights the importance of allowing for drying of the samples after immersion tests, to determine whether extraction occurred which was concealed by swelling. Similar effects were also observed for neoprene with ndecane and 1-heptanol where the high mass uptake observed at day 0 was completely reversed to reveal a significant mass loss. The significant mass gain of neoprene with the inorganic PCMs was also partially reversed but the final value remained positive and nonnegligible (4-12%).  Table 7 summarizes the results of the investigations conducted in this project and attempts to evaluate the PCMs with regard to their suitability as storage media in refrigeration-related latent heat storage units. All PCMs have weaker and stronger areas of performance and no ideal candidate exists. However, some candidates do stand out. The  On day 0 of drying, neoprene showed a very high mass uptake of the organic PCMs (118-431%) and a significant uptake of the inorganic ones (15-24%). It is therefore considered unsuitable for use in storage units especially with organic PCMs, as it would experience very high swelling. Butyl rubber showed a significant uptake of n-decane and propionic acid (38-85.5%) and it is therefore considered unsuitable for utilization with these PCMs. n-Decane also showed high migration in polypropylene, corresponding to a 34% mass gain of the PP sample, and therefore this combination of this polymer and PCM is also not advisable. The rest of the polymer/PCM combinations exhibited a low mass variation and are considered promising combinations for storage applications. Figure 7 (right) presents the mass variations of the same samples, 7 days after they were extracted from the immersion tests. Interestingly, the butyl rubber samples showed significant mass loss with respect to the original value (>4%) for the immersion test with all organic PCMs. This indicates removal of soluble constituents from the butyl rubber which could be observed at day 7, after the uptake of PCM in the polymeric samples has been reversed. This mass loss is considered significant and should be further investigated to determine its effects on the durability of the polymer. This effect highlights the importance of allowing for drying of the samples after immersion tests, to determine whether extraction occurred which was concealed by swelling. Similar effects were also observed for neoprene with n-decane and 1-heptanol where the high mass uptake observed at day 0 was completely reversed to reveal a significant mass loss. The significant mass gain of neoprene with the inorganic PCMs was also partially reversed but the final value remained positive and non-negligible (4-12%). Table 7 summarizes the results of the investigations conducted in this project and attempts to evaluate the PCMs with regard to their suitability as storage media in refrigerationrelated latent heat storage units. All PCMs have weaker and stronger areas of performance and no ideal candidate exists. However, some candidates do stand out. The NaCl/water mixture presented the highest volumetric energy density, more than 30% higher than the second best candidate (Al(NO 3 ) 3 /water), justifying its popularity in existing literature. Its biggest drawbacks are the relatively high degree of supercooling and the fact that pitting corrosion was observed in the immersion tests with aluminum. The former is expected to become less prominent when the PCM is utilized at higher, application-relevant volumes and, if necessary, it can be further reduced by addition of nucleating agents. The latter could lead to the exclusion of any aluminum components from the storage unit that could come in contact with the PCM. This is of course particularly relevant to heat exchangers, where aluminum is often the material of choice because of its low price, low weight, and good thermal conductivity. To circumvent this issue, other metals could be used, such as the more expensive and more thermally conducting copper. For lower temperature applications, Al(NO 3 ) 3 /water 30.5 wt.% showed a high volumetric energy density but also a fairly high degree of supercooling and incompatibility with both aluminum and copper, which makes its implementation in a latent heat storage unit challenging. Further, n-decane presents a more straightforward candidate for this temperature range. With 41.4 kWh/m 3 , n-decane has a good energy density with a low degree of supercooling and very good compatibility with all materials except neoprene. As neoprene can be easily omitted or substituted in the storage unit, the lack of compatibility is not considered critical. Despite a higher degree of supercooling, propionic acid could also be a good organic alternative for refrigeration processes needing a slightly higher melting point, however the use of copper should be avoided in this case. Additionally, 1-heptanol is the least useful investigated candidate, with a relatively low measured energy density.