Eutectic and hypereutectic Cu–Ge alloys were selected to ensure a small volume change during phase change, liquefaction temperatures up to 800°C, and high thermal storage capacity because of the excess Ge content. Generally, most alloys expand thermally with increasing temperature. However, Ge may thermally shrink at the eutectic temperature owing to the nature of the abnormal liquid when the phase changes from solid to liquid. In terms of volume change and latent heat storage, this indicates that excess Ge can alleviate the thermal stress of the PCM container/capsulation during the solid–liquid phase transition in the charge–discharge mode; the excess Ge can contribute to enhancing the latent heat in a narrow temperature range. Hence, the potential of the Cu–Ge alloy was examined as a PCM thermal storage material.

Figure 1 shows the phase diagram of the Cu–Ge alloy. A phase diagram has been previously reported for binary alloys (Okamoto, 2010). Based on the storage temperature to 800°C, the present study selected eutectic (60 wt% Cu–40 wt% Ge, CuGe40) and hypereutectic (50 wt% Cu–50 wt% Ge, CuGe50) (40 wt% Cu–60 wt% Ge, CuGe60) mixtures of the Cu–Ge alloy.

The eutectic mixture corresponded to the CuGe40 = ε 2 phase + (Ge) and behaved in a reversible solid/liquid phase change at T = 644°C, and a solid/solid phase change between the ε 2 and ε 1 phases at T = 614°C. The hypereutectic alloys of CuGe50 and CuGe60 corresponded to a mixture of eutectic mixture and excess Ge solid, CuGe50, and CuGe60 = eutectic mixture + (Ge), and acted in a manner similar to the eutectic mixture at T = 644 °C. Subsequently, they continued to melting/solidification (latent heat storage) of the primary crystal (Ge) and sensible heat storage of liquid and solid phases to liquefaction temperatures of T = 644–765°C. Thus, the eutectic and hypereutectic Cu–Ge alloys exhibit a high latent heat capacity across the range of 614–765°C.

From the viewpoint of volume change during phase change, a hypereutectic Cu–Ge alloy contains more Ge compared to that of the eutectic composition, thereby indicating that the volume change of the PCM can be controlled at the phase change (solid/liquid phase). The possibility of volume change by varying the chemical composition of the Cu–Ge alloy was also investigated in this study.

The Cu–Ge alloys were synthesized from reagents (high-purity materials, KOJUNDO CHEMICAL LABORATORY CO, LTD.) in a laboratory. A fine powder of copper (purity 99.9%, 180 µm) and germanium (4N, 300 µm) were weighed and mixed in an alumina boat. The alumina boat containing the powders was heated at 1,400°C for 2 h in an Ar stream of 1 dm³/min at a normal state to prevent oxidation during the alloying process, and to melt the powder and form ingots of Cu–Ge alloy. The ingots of Cu–Ge alloy were subjected to homogeneous heat treatment at 800–900°C under vacuum. The ingot was cut into pellets (diameter <25 mm and thickness <10 mm) and chemically washed and cleaned to remove oil and dust from the surface. The identification of the solid phase and structural characterization of the Cu–Ge alloy were carried out by X-ray diffraction (XRD) (D2Phaser, Burker) using CuKα radiation (λ = 0.15418 nm) (30 kV-10 mA) at room temperature. The specimens for XRD measurement were prepared by utilizing a mounting press on phenolic resin to encapsulate the alloy and adding a layer of carbon conductive filler. The surface of the alloy was polished using SiC paper and then mounted on a specimen holder ring. Diffraction data were collected in the angular range corresponding to 10° < 2θ < 80° with a 0.02° step size and a 1-s step interval. The crystalline phases were identified by comparison with standard reference patterns (Inorganic Crystal Structure Database, ICSD) (ICSD, 2020) and the Crystallography Open Database (COD). The lattice cell parameters of the refined structures of the synthesized solid phase were evaluated via Rietveld refinement of the structure models by using the pattern fitting method of the FullProf package.

There is a paucity of studies on the values of the specific heat (C _p) based on the temperature and thermophysical properties of the Cu–Ge alloy. The phase transition temperatures and enthalpy of fusion for Cu–Ge alloys with a wide range of chemical compositions have been examined (Zhai et al., 2012). In this study, the melting, solidification, and eutectic temperatures, enthalpies, and specific heat capacity of the alloys were measured using differential scanning calorimetry [DSC, NETZSCH DSC 404 F3 Pegasus^® manufactured by NETZSCH Co. Ltd., temperature resolution of ±0.0025 × |t| °C)] under an Ar flow of 100 ml/min with a heating and cooling rate of 10°C/min between room temperature and 900°C. Heating/cooling cycles in DSC were repeated three times under an Ar stream. From these measurements, the temperature dependence of sensible heat was estimated using the measured C _p values of the Cu–Ge alloy via DSC.

The thermal volume change of the Cu–Ge alloys was measured via thermomechanical analysis (TMA, NETZSCH 4000 SE, temperature resolution of ±0.5°C). To measure the thermal expansion/shrinkage when the phase changes between solid and liquid phases at high temperature, we designed and customized a sample holder to put the sample in the TMA equipment. The sample holder comprised a hollow cylinder (inner diameter of 6.5 mm, outer diameter, 10.5 mm, length, 22 mm) and pistons (diameter, 6.4 mm, length, 8 mm). A test sample of the Cu–Ge alloy (diameter, 6.2 mm, length, 7.5 mm) was prepared in a laboratory. The test sample was placed in the customized sample holder (inner diameter of 6.5 mm, length, 22 mm), and the test unit was placed in the TMA equipment. The coefficient of thermal expansion/shrinkage based on the temperatures and phase change between solid and liquid phases at specific temperatures was monitored under an N₂ flow of 100 ml/min with a heating and cooling rate of 5°C/min. The density of the Cu–Ge alloys was measured at 25°C using the Archimedes method. The temperature dependence of the density of the Cu–Ge alloys was determined using TMA data.

Thermal diffusivity of the Cu–Ge alloys was measured using a NETZSCH LFA-467HT hyper flash analyzer. The test sample (diameter of 10 mm and thickness of 1.14 mm) was prepared in a laboratory. The test sample was placed in a sapphire container, and the test unit was placed in the LFA equipment. Thermal diffusivity measurements were performed with Xe flashlight (pulse width 600 μs, five shots) at 250 V in an Ar stream at 20–900°C to avoid oxidation of the test sample. The thermal conductivities of the Cu–Ge alloys were calculated using these data.

The thermal response of the charge/discharge modes was tested for the eutectic Cu–Ge alloy (60 wt% Cu-40 wt%Ge) as follows: The experimental arrangement for the thermal response tests is shown in Figure 2. A stainless-steel (SUS310S) test container (length, 200 mm; inner and outer diameters, 93.6 and 101.6 mm, respectively), was used for the thermal charge/discharge performance tests. The PCM container (length, 35 mm; inner diameter, 30 mm, and thickness, 10 mm) was composed of graphitic carbon in vacuum, and stainless steel (SUS 310S) in an air-flow atmosphere and was placed inside the test container. The Cu–Ge alloy was packed into a PCM container. 60 g of the eutectic Cu–Ge alloy was loaded into the PCM container (Figure 2B). For the charge mode, the electric heater was controlled at a constant heating rate of 4–6°C/min (1-9 cycles) and 8–12°C/min (10–20 cycles), to a temperature of 850°C, exceeding the eutectic temperature, based on the chemical composition of the Cu–Ge alloy. The temperature variation of the test container with an endothermic phase change was measured under a controlled heating rate. The temperature was maintained for 60 min to homogenize the melt alloy. Subsequently, the test container was subjected to the discharge mode. For discharge, the electric heater was controlled at different cooling rates, ranging within 4–6°C/min (1–9 cycles) and 4–10°C/min (10–20 cycles), to a temperature of 250°C. Temperature variation of the test container with an exothermic phase change was measured under a controlled cooling rate. The charge and discharge modes for the Cu–Ge alloy were repeated in a vacuum and air atmosphere to evaluate the temperature conformity and repeatability of the phase diagram and the effects on the reproducibility of both modes under different atmospheres and heating/cooling rates. The thermal responses were evaluated during heating and cooling (dT/dt vs. time). The PCM temperature was measured relative to time using a type K thermocouple (temperature resolution of ±0.0075 × |t| °C), which was directly inserted into the crucible inside the test container.

After the thermal response tests, the oxidation state, metallographic structure, and extent of elemental distribution of the Cu–Ge alloy were observed and evaluated. Each element in the Cu–Ge alloy was analyzed using an electron probe microanalyzer (EPMA, Shimadzu EPMA-1610) equipped with a wavelength dispersive X-ray spectrometer (WDS, relative error of <1%) operating at an acceleration voltage of 15 kV, beam current of 200 mA, beam size of 1 μm, step size of 0.5 μm, and sampling time of 0.1 s. The samples for the microstructural analyses were prepared via mechanical grinding and polishing using resin bonded diamond grinding discs (Struers, MD-Piano 1,200, 2,000, and 4,000) to obtain a mirror-like finish prior to EPMA analysis. The chemical compositions of the oxidized outer layer and non-oxidized inner layer were quantitatively measured using EPMA analysis. The average chemical compositions of the oxidized and non-oxidized layers were estimated to evaluate the variation in chemical composition via oxidation.

A chemical compatibility test between the Cu–Ge alloy and candidate materials of the PCM container at high temperature was performed. Stainless steel (SUS 310S, rod-shape), alumina (tube-shape), Inconel625 (rod-shape) and silicon carbide (SiC, powder) were selected as candidate materials. A piece of the Cu–Ge alloy (10 g) was placed together with a PCM container candidate material into a round-bottom-shaped tanman tube (alumina 99.99%, inner diameter of 8 mm and length 25 mm) and heated in vacuum at 800°C for 2 h to pre-melt the Cu–Ge alloy in contact with the candidate materials. A test vessel (length, 120 mm; height, 60 mm) made of stainless steel was used to hold the tanman tube. The tanman tubes were vertically arranged in a test vessel (length, 120 mm; height, 60 mm) within a glovebox maintained in an Ar atmosphere to avoid oxygen contamination, which was then bolted and sealed using an elastic airtight packing (O-ring) made from copper. The test vessel was heated for 720 h at 800°C in a muffle furnace. After the heat treatment, the tanman tubes were removed from the test vessel and cut parallel to the horizontal axis. A piece of test sample was embedded in epoxy resin to observe and evaluate the cross section of the alloy and candidate material of the PCM container. To evaluate the chemical compatibility and the extent of element distribution of the alloy, each of the constituent elements contained in the alloy and the candidate materials were analyzed using an electron probe microanalyzer (EPMA, Shimadzu EPMA-1610) equipped with a wavelength dispersive X-ray spectrometer operating at an acceleration voltage of 15 kV, a beam current of 200 mA, beam size of 2 μm, step size of 2 μm, and sampling time of 30 ms. The samples for the EPMA analysis were mechanically ground and polished using resin bonded diamond grinding discs (Struers, MD-Piano 1,200, 2,000, and 4,000) to obtain a mirror-like finish before EPMA analysis.

Figure 3 shows XRD patterns of the synthesized CuGe40, CuGe50, and CuGe60 alloys at room temperature in the 2θ range. Blue vertical bars below the XRD pattern indicate the peak positions of the Cu–Ge alloys. The peak positions were compared with XRD data of some phases obtained from the ICSD and COD standard databases, and the phases were identified. A series of peaks, denoted by green vertical bars below the XRD patterns, correspond to the Cu₃Ge phase (ICSD-86007). The Cu₃Ge phase was equivalent to ε 1 phase at room temperature and high-temperature stable ε 2 phase in the phase diagram (Figure 1). Neither phases are line compounds, but small chemical composition range (solid-solution). The other series of peaks, denoted by orange vertical bars below the XRD patterns, correspond to the Ge phase (COD-96–900–8568). The phase was equivalent to Ge solid-solution phase in the phase diagram (Figure 1) and observed for all Cu–Ge alloys at room temperature. Other peaks were not observed in the pattern. All diffraction peaks were assigned as ε 1 phase and Ge solid-solution phases, and lattice cell parameters of solid phases were evaluated via Rietveld refinement of the structure models and the results are listed in Table 1. All peaks assigned by ε 1 phase were identified as an orthorhombic unit cell [space group P m n m (59)] while those specified by Ge solid-solution phase were determined as a cubic unit cell [space group F d −3 m (227)]. The lattice parameters for Ge solid-solution phase were very close to that [a = 5.657 (3) Å] obtained from the database. The lattice parameters for ε 1 phase were identified and compared to those (a = 5.272 Å, b = 4.204 Å, c = 4.578 Å). Small deviations for all Cu–Ge alloys were observed. This is because of solid-solution with a very narrow gap of chemical composition. The results indicated that the alloys were successfully synthesized without impurities as per the phase diagram of the Cu–Ge alloy (Okamoto, 2010).

The temperature dependence of specific heat capacities, C _p , and thermal expansion/contraction fraction, α , at high temperatures have not been reported to-date. In the present study, the thermophysical properties were measured and reported for all alloys in the temperature range 100–900°C. The C _p data were utilized to estimate the sensible heat and evaluate the thermal storage performance. Figure 4A shows specific heat capacity and (b) volume change as a function of temperature for all the alloys. The approximation formula of the solid phase was estimated from the secondary fitting of the measured C _p values, while that of the liquid phase was calculated from the linear fit to obtain a C _p-T correlation in the following forms:

CuGe40 alloy:

Solid phase C p = 3.43 × 10 − 7 T 2 − 8.58 × 10 − 5 T + 4.37 × 10 − 1   ( 100 − 600 ° C ) (1) Liquid phase C p = 0.5223 + 6.794 × 10 − 6 ( T − 700 )     ( 700 − 920 ° C ) (2) CuGe50 alloy:

Solid phase C p = 1.02 × 10 − 6 T 2 + 1.28 × 10 − 3 T + 2.84 × 10 − 1   ( 100 − 600 ° C ) (3) Liquid phase C p = 0.6840 + 7.536 × 10 − 4 ( T − 730 )     ( 730 − 875 ° C ) (4) CuGe60 alloy:

Solid phase C p = 3.69 × 10 − 7 T 2 + 9.34 × 10 − 4 T + 3.35 × 10 − 1   ( 100 − 600 ° C ) (5) Liquid phase C p = 0.7960 + 5.444 × 10 − 4 ( T − 780 )   ( 780 − 880 ° C ) (6) where C _p has units J/(g°C), and T is in °C. The results shown in Figure 4 were the third dataset, when the experiments were repeated three times. The standard deviations were as follows: 2.16 × 10 − 3 @500°C and 4.00 × 10 − 3 @800°C for CuGe40, 4.47 × 10 − 3 @500°C, and 4.90 × 10 − 3 @800°C for CuGe50, 4.97 × 10 − 3 @500°C and 8.87 × 10 − 3 @800°C for CuGe60. The coefficients of determination for Eqs 1, 3, 5 were R ² = 0.998, 0.996, and 0.998, respectively. All equations for the solid phase exhibited a large positive slope for the C _p-T correlation. However, all equations for the liquid phase exhibited a very weak slope of the C _p-T correlation. These results suggest that the heat capacity of the Cu-Ge eutectic and hypereutectic alloys has a weak temperature dependency in the liquid state. The coefficients of determination for Eqs 2, 4, 6 were R ² = 0.884, 0.872, and 0.667, respectively. The values for R ² were smaller for the liquid phase than for the solid phase. Thus, Eqs 2, 4, 6 should be applicable in the assigned temperature range. The measured C _p for both phases were utilized to estimate the sensible heat of the solid and liquid phases.

Figure 4B shows the temperature dependence of the linear thermal expansion/contraction   α for all alloys. The standard deviations were as follows: 8.12 × 10 − 2 @500°C and 5.95 × 10 − 2 @800°C for CuGe40, 3.15 × 10 − 2 @500°C, and 7.68 × 10 − 2 @800°C for CuGe50, 7.52 × 10 − 3 @500°C, and 5.59 × 10 − 2 @800°C for CuGe60. The fraction of thermal expansion/contraction was measured as a standard point at 25°C. The slope of the solid phase at ∼ 600°C is positive for all alloys, and the degree of thermal expansion was reduced with increasing Ge content. The values of thermal expansion at 600°C were 0.47% for CuGe40, 0.25% for CuGe50, and 0.13% for CuGe60. These results suggest that the Ge solid-solution phase has a small expansion coefficient compared to the ε 1 phase in the solid state. At T = 614°C, all alloys expanded to 0.93% for CuGe40, 0.72% for CuGe50, and 0.54% for CuGe60. This is due to the solid phase transition from the ε 1 to ε 2 phase at this temperature. Subsequently, CuGe40 apparently shrank to 0.36% at a eutectic temperature of T = 644°C. This result suggests that the Ge solid-solution phase has a negative coefficient of thermal expansion when the solid phase melted. For CuGe50, a thermal shrinkage of −1.68% was observed at a liquefaction temperature of T = 705°C. A thermal shrinkage of −3.79% was apparent at a liquefaction temperature of T = 765°C for CuGe60. Thermal shrinkage increased with increasing Ge content, which indicates that the Ge solid-solution phase shrinks upon melting at T = 644–765°C. Finally, all the alloys exhibited increasing thermal expansion over the liquefaction temperatures. These results indicate that the liquid phase thermally expanded with increasing temperature. The thermal expansion for all alloys is capable of controlling to +1.85 to −2.55% based on the chemical composition at T = 900°C.

Figure 5A shows the temperature dependence of the apparent density of the CuGe alloys. The apparent density was calculated using the third dataset of thermal expansion/contraction. The apparent density of all alloys at 25°C was measured using Archimedes’ principle and estimated from the volume change data at various temperatures (Figure 4B). The apparent density of CuGe40 was in the range ρ = 7 .50 − 7 .45   g / c m 3 for the solid phase at T = 25–600°C (Figure 5A). When the ε 1 phase transformed to the ε 2 phase at T = 614°C, the apparent density decreased slightly. Subsequently, density increased owing to melting of the eutectic mixture at T = 644°C. The density was ρ = 7 .37   g / c m 3 for the liquid phase at T = 900°C. The rate of density change for CuGe40 was estimated to be 1.73% between T = 25–900 °C. For CuGe50, the apparent density ranged between 7.11 and 7.14 g/cm³ for the solid phase at T = 25–600°C. In the case of the CuGe60, the apparent density ranged between 6.76 and 6.78 g/cm³ for the solid phase at T = 25–600°C. Owing to the hypereutectic mixture, the density increased to a liquefaction temperature of T = 705°C for CuGe50 and a liquefaction temperature of T = 765°C for CuGe60. The rates of density change for CuGe50 and 60 were estimated to be +0.56 and +2.66% between T = 25–900°C, respectively. The results of very small density changes as a function of temperature indicate that the PCM is capable of loading rate over 95% when the PCM alloys are packed into a thermal storage container filled in a storage tank/capsule. Thus, the thermal storage capacity per storage volume can be enhanced by using PCM alloys.

In theory, the density of liquid metal at high temperatures can be described by a linear equation, as demonstrated by Cailletet and Mathias (Cailletet and Mathias, 1886). Thus, the measured density data were fitted to the following linear equations:

CuGe40 alloy:   ρ = − 6.129 × 10 − 4 T + 7.918 ,   R 2 = 0.9997   at   T = 650 − 900°C (7) CuGe50 alloy:   ρ = − 4.085 × 10 − 4 T + 7.548 ,   R 2 = 0.9997   at   T = 700 − 900°C (8) CuGe60 alloy:   ρ = − 6.156 × 10 − 4 T + 7.503 ,   R 2 = 0.9999   at   T = 760 − 900°C (9) where temperature T is in °C.

Figure 5B shows the temperature dependence of thermal diffusivity, λ and conductivity κ for the CuGe40 alloy. The standard deviations of the measured thermal diffusivity were 1.26 × 10 − 2 @500°C and 1.50 × 10 − 1 @800°C for CuGe40. The values of thermal diffusivity decreased with increasing temperature in the solid phase at temperatures of 25–600°C and decreased to approximately 25% of the original value in the liquid phase at temperatures of 675–900°C when the alloy melted. Thermal diffusivity, λ , is a material-specific property that characterizes self-diffuse thermal energy; this value represents how quickly the material responds to temperature changes. Thermal conductivity, κ , is defined as the amount of energy transferred as heat that flows through the mass in response to a temperature gradient. Thus, the thermal conductivity, κ , was calculated from values of density ρ, specific heat capacity, C _p, and thermal diffusivity λ according to Eq. 10. κ = ρ × C p × λ  (10) The temperature dependence of the calculated thermal conductivity is shown in Figure 5B. The thermal conductivity at a low-temperature (T < 100°C) for the solid phase of CuGe40 was higher than that for pure germanium (59.9 W/(°C m) @27°C) (AIST, 2019) and much lower than that of pure copper (386 W/(°C m) @ 27°C) (AIST, 2019) and comparable to that at high temperatures (500–600 °C). A main reason for this is the coexistence of the ε 1 phase and the Ge solid-solution phase. This result indicates that the ε 1 phase is more thermally conductive than the Ge solid-solution phase at low temperatures, thereby leading to an increase in the thermal conductivity of the solid mixture. In addition, the solid mixture forms a periodic lamella microstructure, as seen in Chemical Compatibility Test at High-Temperate, which appears in the eutectic mixture of the ε 1 and Ge solid-solution phases and involves a homogeneous nucleation-growth process. At high temperatures (700–900°C), the thermal conductivity of 15–8 W/(°C m) in a liquid mixture was much lower than that in a solid mixture. The trend of relatively low thermal conductivity in the liquid phase has been observed in other alloys (Blanco-Rodríguez et al., 2014; Risueño et al., 2017).

The following equations give the fitted second-order polynomial for these properties in the liquid phase: λ = 2.59 × 10 − 5 T 2 − 4.98 × 10 − 2 T + 2.60 × 10 1   ,   R 2 = 0.999   at   T = 650 − 900°C (11) κ = 1.08 × 10 − 4 T 2 − 2.11 × 10 − 1 T + 1.10 × 10 2   ,   R 2 = 0.999   at   T = 650 − 900°C (12)

Third-generation (Gen3) CSP technology has a new target for the temperatures of HTF (T > 700°C) to ensure higher energy efficiency using supercritical CO₂ thermal cycles. Chloride eutectic molten salts are suitable for meeting some important thermophysical properties, including: 1) a melting point that is as low as possible; 2) a boiling point at least 800°C or above; 3) thermophysical properties that are acceptable for convective heat transfer and thermal storage; 4) low corrosion to metal pipes and containers at high temperatures; and 5) low cost. The University of Arizona selected and tested binary eutectic KCl/MgCl₂ molten salt that can be used as HTF and sensible thermal storage for next-generation CSP plants (Xu et al., 2018). Researchers at the National Renewable Energy Laboratory (NREL) of the United States reported a new ternary eutectic chloride salt comprising MgCl₂-KCl-NaCl as a third third-generation high-temperature HTF and TES in the community of Gen3-CSP (Zhao and Vidal, 2020). In the present study, thermophysical properties of binary and ternary chloride molten salts were compared to the Cu–Ge alloys on the basis of assumption that the molten salts was used as PCMs.

The latent heat storage capacities and dependence of the specific heat, C _p, in the range 100–800°C were measured for all alloys, and the sensible heat was estimated to evaluate the thermal storage performance (Figure 4A). Figure 6 shows (a) heat storage density per unit weight and (b) thermal storage capacity per unit volume for the Cu–Ge alloys. The measured C _p for the solid and liquid phases were utilized to estimate the sensible heat capacities of the solid and liquid phases, respectively. The sensible heat capacity of the solid phase increased with temperature in the range 100–614°C (Figure 6A). In addition, the slope increased with increasing Ge content in the Cu–Ge alloys. The average specific heat capacity of Cu [0.386 J/(g °C)] is higher than that of Ge [0.310 J/(g °C)] in the temperature range 0–100°C (Kinzoku data book, 2018). Thus, these results indicate that the specific heat capacity of Ge has a relatively strong sensitivity at high temperatures. Small discontinuous changes are the sum of sensible heat and latent heat released/stored in the solid/solid phase transition ( ε 1 ⇔ ε 2 phase) at a temperature of 614°C. The latent heat were estimated as 41.8, 27.3, and 23.2 kJ/kg for CuGe40, CuGe50, and CuGe60, respectively. The value of latent heat for CuGe40 was estimated by reference data (28.4 kJ/kg) (Zhai et al., 2012). Strong discontinuous changes are the sum of sensible heat and latent heat released/stored in the solid/liquid phase transition (eutectic mixture of ε 2 + ( Ge ) ⇔ liquid phase) at a temperature of 644 °C. The latent heat were estimated as 232.8, 198.2, and 163.5 kJ/kg for CuGe40, CuGe50, and CuGe60, respectively. The value of latent heat for CuGe40 is comparable to reference data (169.7 kJ/kg) (Zhai et al., 2012). Additionally, the heat storage density of CuGe50 increased in the temperature range 644–705°C owing to the latent heat from the primary Ge and sensible heat from the eutectic mixture, while the heat storage density of CuGe60 is enhanced in the temperature range 644–765°C. Beyond the liquefaction temperature, which depends on chemical composition, the sensible heat from the liquid phase contributes to increases in the heat storage density of the Cu–Ge alloys. An advantage of the hypereutectic chemical composition (CuGe50 and CuGe60) is that it reinforces the total storage at high temperatures over the eutectic temperature. An advantage of the eutectic mixture (CuGe40) is that it behaves like a simple substance performing solid/liquid phase transition at a constant temperature.

The thermophysical properties of the Cu–Ge alloys were compared with those of the binary chloride molten salt (Table 2). The heat storage densities per unit weight for the Cu–Ge alloys were compared to those of the binary chloride molten salt (Xu et al., 2018), and are plotted in Figure 6B. For the ternary molten salt, the values of C _p in the solid state have not been previously reported (Zhao, 2020); thus, the data were not plotted. The accumulated storage density at 700°C was greater for the chloride molten salt than for all alloys (Figure 6A). In addition, the liquid phase of the molten salt has a larger storage capacity compared to all alloys. It is very promising as an HTF and liquid TES in next-generation CSP technologies. However, the accumulated sensible heat capacities of solid phase for all alloys was higher than that of the chloride molten salts (Figure 6B) due to the high density of the alloys (molten salt of 1.89 g/cm³ (extrapolation value at 25°C); Cu–Ge alloys of 6.77–7.50 g/cm³ (25°C). The latent heat capacities of all alloys at the eutectic temperature were also superior to those of chloride molten salts. In addition, all alloys exhibited a high thermal conductivity in the solid/liquid phases and little density change at the solid/liquid transition. These results indicate the potential of all alloys as PCMs, based on the limitations of the usable volume in an LTES system. In the ongoing Gen3 liquid pathway project, two options are considered suitable for reaching high temperatures: 1) direct heating of a chloride molten salt; and 2) indirect heating via liquid sodium and an associated sodium‒salt heat exchanger (Energy Efficiency and Renewable Energy, 2021). Both options use a two-tank system comprising a hot tank and cold tank for chloride molten salt storage. To reduce the storage cost of constituent materials for large quantities of thermal energy storage, 1-tank thermocline system, which is a storage mixture of low-cost sensible material (for example, rock) and latent PCM filler (for example, capsulated alloy), may be a promising technology in view of the limitations of the useable volume for TES. The volumetric capacity of heat storage is a key parameter for next-generation CSP technologies.

When the alloys are installed as latent heat storage (PCM) into the TES of a CSP plant, the alloy is subjected to numerous day–night cycles, and the evaluation of repeatability and compatibility is necessary. In the present study, short-term reliability (repeatability of the charge/discharge performance and compatibility with the PCM container) was initially evaluated via 20 cycle tests using a few dozen Gram samples (Figure 7). From a thermodynamic viewpoint, CuGe40 must behave reproducibly based on the phase diagram (Figure 1) during the thermal response test if it does not chemically react with the container and stream gas. From a kinetics viewpoint, for the phase transition between solid and liquid phases, it is desirable that charge/discharge responds quickly at the eutectic temperature assigned in the phase diagram.

Figure 7 shows dT/dt vs. the PCM temperature T for the CuGe40 in an air stream and a vacuum. The thermal response is influenced by heating/cooling rates, heat transfer, gas stream in the PCM container, and container material. Thus, the main purpose is to examine thermal response and repeatability of the CuGe40 placed in a crucible in an air stream. The crucible was made of SUS304 stainless steel with excellent thermal conductivity and oxygen resistance. To examine the impact of oxygen in the gas stream on the charge/discharge performance, the test was performed using ambient air. In addition, tests in vacuum were conducted to compare the impact of thermal conduction and radiation without convective heat transfer from the external heater into the PCM. Strong endothermic peaks for all cycles were observed immediately after, at temperatures of 614 and 644°C during the heat charge mode (Figure 7A). The peak temperatures were in agreement with the ε 1 ⇔ ε 2 phase transition (614°C) and eutectic (644°C) temperatures of the CuGe40 alloy in the phase diagram (Figure 1). The endothermic peak was largest during the first cycle and relative to small during the 2nd–20th cycles. These results indicate that the surface of the CuGe40 alloy was oxidized by the air stream during the first cycle, but the internal fraction of the alloy functioned as a PCM throughout the 2nd–20th cycles. In addition, the charge performance for the CuGe40 alloy after the oxidation were obtained with high repeatability in static (constant) heating mode. A large peak was observed for all cycles at 681°C given the termination of the latent heat storage. It was caused via rapid heating of the CuGe40 alloy, which retained relatively low temperatures due to the absorption of latent heat, compared to the heater temperature, which automatically increased at a constant rate. The variation in the dT/dt profile in the charge mode indicated that the latent heat storage of the CuGe40 alloy under different heating rates was maintained, with good conformity to the phase diagram, with no deterioration of performance in the PCM container.

During heat discharge, a strong peak consistently appeared at 644°C, and a strong peak appeared at 614°C (Figure 7B). The exothermic behaviors for all cycles corresponded to the solidification of the eutectic mixture from the melting alloy and the solid/solid phase transition ( ε 1 ⇔ ε 2 ). As observed during heat charge mode (Figure 7A), the peak intensity of the two exothermic peaks was larger for the first cycle than for the 2nd–20th cycles. In addition, the exothermic peaks for the 2nd–20th cycles were shifted to lower temperatures. The discharge performance for the CuGe40 alloy was observed with good repeatability during the 2nd–20th cycles. These results indicate that an oxidation layer formed on the surface of the CuGe40 alloy caused a late response of exothermic heat discharge during the 2nd–20th cycles. The charge or discharge of latent heat from a eutectic mixture and solid/solid phase transition occurs under quasi-thermodynamic equilibrium conditions at a fixed atmospheric pressure (Figures 7A,B). Thus, the experimental results for the dT/dt profiles indicated that the melting or solidification of the eutectic mixture and the solid/solid phase transition consistently occurred under different cooling rates, without supercooling.

Figure 7C shows the dT/dt profiles plotted relative to the PCM temperature measured under vacuum. The cyclic thermal response was examined using a PCM container composed of graphitic carbon. Additionally, heat charging was imposed at heating rates of 5–6°C/min for the first ten cycles, but at 8–10°C/min for the remaining ten cycles. The thermal response of the CuGe40 alloy in a vacuum is governed by thermal conduction and radiation from the heater into the alloy. An endothermic very weak peak due to solid/solid phase transition for all cycles was observed at an onset temperature of 633°C (Figure 7C), and the peak temperature was shifted to a higher temperature when in an air stream (Figure 7A). Similarly, a strong endothermic peak, due to the melting of the eutectic mixture, was observed at temperatures of 651–663°C. This is because of deficient and inhomogeneous heat transfer without convection in a vacuum. The endothermic behavior in the dT/dt profiles was almost the same for different heating rates. The repeatability of charge performance under vacuum during the 2nd-20th cycles was inferior to in the air stream.

During heat-discharge (Figure 7D), the cooling rate was set to 4–5°C/min for the initial ten cycles and dynamically changed between 6 and 9°C/min for the remaining ten cycles. A very strong exothermic peak, due to the solidification of the eutectic mixture from the melting alloy, was observed for all cycles, with an onset temperature of 644°C. However, an exothermic peak due to the solid-state phase transition appeared at an onset temperature of 610–598°C. The reason for this fluctuation is the cooling process in vacuum, which is governed by the limited heat transfer. Thus, the heat stored in the alloy was not quickly released into the atmosphere. This leads to a delay and fluctuation (poor repeatability) in the onset temperature of the exothermic process. The impact on the cooling rate was not clearly observed during discharge under vacuum. The thermal response of the exothermic peaks was consistently observed under fluctuating onset and termination temperatures for the solid-state phase transition and solidification of the eutectic mixture for all cycles. The CuGe alloy behaved consistently, based on the thermodynamic equilibrium in a gas stream and quickly responded under the limited heat transfer without the effect of alloy oxidation in vacuum.

To evaluate the effect of oxidation during the thermal response test in an air stream, and chemical compatibility between the stainless steel vessel and CuGe50 alloy (sample), the microscopic distribution of the sample and vessel elements was examined after the test. Figure 8 shows the backscattered electron image (BEI) micrograph of the cross-sectional surface of the sample and elemental mapping images of Ge, Cu, Fe, O, Ni, and Cr. These analytical elements were selected from the main constituents of the sample and the vessel. The BEI image consists of a light gray area (right side of the image), a dark gray area (upper-left side of the image), and a black area (lower left side of the image). The black area corresponds to the epoxy resin used to mount the sample. The light gray area represents the internal microstructure of the alloy without oxidation, while the dark gray area is an oxidation layer in which the alloy was externally covered. The BEI image indicates that an oxidation layer was formed on the surface of the alloy during the heat charge-discharge process in air. In the mapping images of the Ge element, the green region is widely spread in the oxidation layer (dark gray area) in the BEI image. In addition, a number of red, yellow and green microscopic spots and needles (Ge solid solution, aggregation of coarsened particles) and a blue region (mainly the ε 1 phase) characteristic of a metallographic eutectic structure, were observed in the internal microstructure of the alloy. These results indicate that the ε 1 phase and Ge solid solution were homogeneously distributed in the eutectic mixture of the alloy.

The blue region of the Ge mapping image corresponds to the brown region of Cu mapping. A number of reddish-brown spots and yellowish-brown areas surrounding the reddish-brown spots appeared in the Cu mapping. Because the reddish spots have a relatively higher Cu content than the yellowish area, the authors consider that the brown region of the Cu mapping is composed of two phases: the reddish spots correspond to the thermodynamically stable ε 1   phase (76.9 %–72.9%Cu) (Okamoto, 2010) and the yellowish area corresponds to the high-temperature stable ε 2   phase (73.95–73.5 at%Cu) (Okamoto, 2010) with slightly lower Cu contents than the ε 1   phase. The ε 2   phase at room temperature can be changed into the ε 1   phase, leading to an increase in the latent heat via the solid-state phase transition.

As seen in the O mapping image, the internal microstructure of the alloy was observed with no oxidation; the formation of the oxidation layer was observed in the dark gray area of the BEI image. The thickness of the oxidation layer was in the range of 100–200 mm (Figure 8A). The chemical composition of the alloy and oxidation layers were quantified by randomly selecting five points by EPMA analysis (Table 3). The average weight ratio of Ge_65.1Cu_5.4O_29.5 for the oxidation layer corresponded to a solid mixture of GeO₂/CuO (or Cu₂O) = 10/1, and that of Cu₆₄Ge₃₆ for the alloy corresponded to a solid mixture of hypoeutectic chemical composition (67 at%Cu-33 at%Ge). These results indicate that Ge was preferentially oxidized compared to Cu in the CuGe50 alloy, resulting in compositional deviation of the alloy. The chemical composition and thickness of the oxidation layer may impact on charge/discharge storage performance and thermal response. The thermal response of the PCM lowered by the formation of the oxidation layer as far as appears in Figures 7A,B. In order to quantitatively estimate the impacts on storage performance under the oxidation layer, a numerical simulation of the PCM covered the oxidation layer and comparison with the experiments are required in some future works.

As seen in the mapping images (Figure 8) of the Fe, Ni, and Cr elements, the internal region of the alloy and the oxidation layer did not chemically react with the stainless steel container under the test conditions. These results indicate that the container (SUS310S) can be used as a PCM container in an air stream during charge-discharge modes.

To use the alloy as a PCM in a thermal storage system, the alloy should be encapsulated by a material having an excellent thermal conductivity, high-temperature resistance, and corrosion resistance at high temperatures. Consequently, the chemical compatibility between the alloys and candidate container materials was tested at high temperatures. The chemical compatibility between the alloys (CuGe40, CuGe50, and CuGe60) and the candidate materials (stainless steel (SUS 310S), alumina, Inconel625, and silicon carbide (SiC)) was examined and evaluated by cross-reactivity at high temperature using EPMA analysis. Figure 9 shows the results of the EPMA analysis for a combination of SUS 310S/CuGe50. A light gray area of Cu-rich and Ge-poor contents, and dark gray areas of Ge, Fe, Ni, and Cr-rich contents, were observed in the backscattered electron image (BEI) micrograph (Figure 9A). In addition, the high Ni content region was locally distributed in the Cu-rich alloy. A black region observed on the left side of the image corresponded to the epoxy resin used to mount the sample. No oxidation of either area appeared in the mapping image of the O element (Figure 9G), which supports the compatibility test without the oxidation effect. These results show that corrosion of the stainless steel into the CuGe50 alloy occurred at high temperatures and SUS310S stainless steel was not chemically compatible with the studied alloy.

Figure 10 shows the results of EPMA analyses for a combination of Inconel625/CuGe40. A vertical array of voids appeared at the center of the BEI image (Figure 10A). This was due to the initial bonded interface of Inconel625/CuGe40, or Kirkendall voids, which generally are an accumulation of atomic vacancies generated by the imbalance in the interdiffusion on the bonded interface. The observation area of the BET image was classified as deep dark gray of Ni-rich content, dark gray area of Cu-rich and Ge-poor contents, and light gray areas of Cr, Ge, and Mo-rich contents. These results show that the Cu and Ge atoms of the alloy diffused into the Inconel layer, and Ni atoms diffused into the CuGe alloy at high-temperatures. Thus, Inconel625 is chemically bonded with the CuGe40 alloy during the test, which is not chemically compatible with the combination at high-temperatures.

Figure 11 shows the results of the EPMA analyses for a combination of alumina/CuGe60/SUS 310S, that is, the chemical compatibility between the tanman-tube crucible made from alumina (purity of 99.99%) and CuGe60/SUS 310S. No dissolution of Al and O from alumina was found in the Cu-rich and Ge, Fe, Ni, and Cr-rich alloys. These results demonstrated that alumina is fully compatible with CuGe60 alloy in a closed inert atmosphere at high temperatures. Figure 12 shows the results of the EPMA analysis for a combination of CuGe50/SiC heated in a tanman-tube crucible. The BEI image demonstrates an interface, where the SiC powder is on the left side and CuGe50 alloy is on the right is clearly observed in the cross-section of the surface of the CuGe alloy and SiC. In addition, Si and C were distributed on the left side without interdiffusion, while Cu and Ge were distributed on the right, without spreading on the other side. After the test, the combined material was easily removed from the tanman crucible. Therefore, no chemical corrosion occurred for the combination of CuGe50/SiC. These results demonstrated that SiC has good compatibility with the chosen alloy and crucible under an inert atmosphere.