Experimental Test of Properties of KCl–MgCl₂ Eutectic Molten Salt for Heat Transfer and Thermal Storage Fluid in Concentrated Solar Power Systems

Concentrated solar thermal power (CSP) systems have an advantage over photovoltaic power generation technologies due to the fact that a CSP system can store solar thermal energy and using it when required at nighttime or under cloudy weather [1]. For conversion of thermal energy to electrical energy in a CSP plant, a high temperature of the heat source is important to the energy conversion efficiency. Therefore, the heat transfer fluid (HTF) in a CSP system plays a critical role. In the present CSP technologies, the highest temperature of heat source is largely restricted by the temperature that a HTF can tolerate. A HTF is used to carry the heat from a solar collector and deliver to the thermal power system; and it is very beneficial if a HTF is also used as a liquid for thermal energy storage [2–5]. Fluids that can be used for heat transfer may include gases, like air, helium, carbon dioxide [6,7], water [8], oils [9,10], or molten salts [11–13]. However, for both effective heat transfer and thermal energy storage at high temperatures, molten salts have been considered as the most liable candidates. This is because that a high temperature HTF needs to hold more heat in a limited space in energy storage system and also being able to transfer heat efficiently, and therefore, the material of HTF must have high density, large heat capacity, low viscosity, and greater thermal conductivity.

This study aims at developing a HTF using chloride eutectic molten salts to meet some important requirement of thermophysical properties including: (1) high operating temperatures of no less than 800 °C, at which the vapor pressure is no larger than 1.0 atm; (2) great thermophysical properties for thermal storage and heat transfer; (3) low cost; and (4) low corrosion for metal pipes and containers.

A heat transfer fluid needs to be chemically very stable at the desired temperatures of application. It is also challenging to find a single material of fluid having both low melting temperature and high boiling point as well as excellent properties for heat transfer and thermal storage [14]. The studies about molten salts for many years have found that mixing different salts to form a eutectic system is a viable approach to lower the melting point and hold the high boiling temperature in the meantime. For example, eutectic nitrate molten salts were employed as HTF which could operate at temperatures as high as 550 °C [15].

In order to achieve the target of a temperature as high as 800 °C for a fluid being able to serve as HTF and thermal storage material in advanced CSP technologies bearing supercritical CO₂ power cycles, a system of chloride salts have been examined and recommended [15,16] due to their favorable thermal and transport properties, chemical stability at high temperatures, relatively low cost, and good availability from natural resources. A comprehensive literature survey has been reported in the authors' previous papers [15–17]. It has been found that MgCl₂ and KCl are two chloride salts candidates with high boiling point (1420 °C and 1412 °C, respectively), and can be available from mining or from the products of seawater desalination.

In this study, MgCl₂–KCl binary eutectic molten salt with molar ratio of 32–68% (mass ratio of 43.4–56.6%) was selected, and its thermophysical properties were experimentally tested, which were expected to meet the requirement of HTF to be used in CSP. The basic properties of MgCl₂ and KCl are shown in Table 1. The theoretical melting point for the eutectic salt of 32 mol % MgCl₂–68 mol % KCl is 430 °C.

The theoretical melting point of eutectic salt mixtures was obtained assuming that the individual salts have 100% purity. However, commercial available salts have more or less impurities, and test for the melting point is needed. To study the thermophysical properties of MgCl₂–KCl eutectic salt, anhydrous salts MgCl₂ (Alfa Aesar, Tewksbury, MA, 99%) and KCl (Alfa Aesar, 99%) were purchased. The two salts were dry mixed inside a dried glove box where the humidity was controlled less than 3%, which was monitored by a psychrometer (Model 69008 by Yellow Jacket, relative humidity range 0–100%). After ground and mixed, the slats were put into a flask (Pyrex, 500 mL) and covered with a lid. Then, the mixture was heated up to 500 °C inside a furnace (Model 3K by Bartlett) and hold for 24 h to make sure all the salts were melted. After cooling down naturally, the molten salts were ground into powder and stored in dried glove box for experimental tests.

The measurements of melting point, heat of fusion, and the specific heat capacity were carried out using a simultaneous differential scanning calorimetry and thermogravimetric analysis (TG) system (Model STA449 F3 made by NETZSCH). The reliability of this system was checked in our previous work for the measurement of eutectic salts NaCl–KCl–ZnCl₂ [18]. The results in that work showed that the difference in melting point and the deviation of the heat of fusion were below ±1 °C and 4.5%, respectively, when compared to data in literature [18].

The curves from TGA signal varying with temperature in the range of 80–710 °C are shown in Fig. 1 for the MgCl₂–KCl molten salt. Ten separate measurements each with fresh salts were conducted. There is an obvious mass loss in the temperature range of 100–200 °C, which is likely resulted in from the vaporization of minor amount of water and other impurities (less than 1.0%) due to the reaction of MgCl₂·KCl·6H₂O = MgCl₂·KCl·2H₂O + 4H₂O(v) at 90–100 °C followed by MgCl₂·KCl·2H₂O = MgCl₂·KCl + 2H₂O(v) and then MgCl₂ ·KCl·2H₂O = KCl + MgOHCl + HCl(g) at 150–200 °C [19,20]. We can also see a small variation of the specific heat capacity caused by this minor mass loss in the same temperature range from Fig. 2. The mass loss was less than 1%, which was reasonable if the purity of constituent salts of 99% is considered. All the TG profiles of ten tests indicate that the mass loss in the MgCl₂–KCl eutectic molten salt in the temperature range of 200–710 °C is very small.

Figure 2 shows the test results of heat capacity against the temperature for eutectic MgCl₂–KCl molten salt. The curves from ten repeated tests are similar. The small “hump” on the C_p–T curve in the temperature range of 100–200 °C is resulted from the vaporization of water. The peaks between 400 °C and 500 °C give us the information about the melting point and heat of fusion. The melting point is determined by the onset of the peak, and the heat of fusion is calculated as the area between the peak and the baseline. Theoretically, the salt gets melted only at the melting temperature, but in the experiment, the melting of the salt takes certain amount of time in a small range of temperature. The $Cp−T$ curves in Fig. 2 indicates that the salt sample can be considered of fully melted at the temperature of 450 °C.

where $Cp$ has a unit of J/g K, and $T$ is in $°C$⁠. Equation (1) shows that the $Cp−T$ correlation has a very small slope, which suggests that the heat capacity of MgCl₂–KCl eutectic salt has a very weak temperature-dependency in the liquid state region. Therefore, Eq. (1) should be applicable in the temperature range from 450 °C to 800 °C.

where $X$ represents $Cp$ at a specified temperature, $N=10$ is the number of tests, and $tN−1,95%$ is student t number, which is $2.262$ at a confidence interval of 95%. Table 2 shows the calculated data at several chosen temperatures, from which we see that the maximum uncertainty is 4.91 × 10⁻² J/g K. Thus, the uncertainty of measurement is bounded by 4.91 × 10⁻² J/g K.

where $ν=N−(m+1)$⁠, and $m$ is the order of the polynomial fit of the data. Here, $m=1$⁠, and $N=2110$ which is because of the ten tested curves, each having 211 data points.

Because the calculated uncertainty of curve fitting u_c is 2.21 × 10⁻³, which is much smaller than the uncertainty of measurement, and thus, the overall uncertainty is still 4.91 × 10⁻² J/g K. The overall uncertainty indicates the upper and lower bound of the data-fitted correlation and curve as shown in Fig. 3.

For melting point and heat of fusion, the averaged value, standard deviation, and uncertainty at a 95% confidence interval of each property were calculated based on Eqs. (2) and (3). As seen in Table 3, the averaged experimental melting temperature is 424.4 °C for the MgCl₂–KCl eutectic salt, where the relative error is 1.3% against the theoretical melting point 430 °C. This difference is caused due to the impurities from constituent salts.

The average value of heat of fusion $ΔHm¯$ is 207.0 kJ/kg from the ten measurements, and the standard deviation $SΔHm¯$ and uncertainty $uΔHm¯$ (defined by Eqs. (2) and (3), respectively) at a probability of 95% are given in Table 4.

Vapor pressure of the molten salt for thermal storage media is an important factor for the operation of a CSP plant. The value of vapor pressure larger than 1.0 atm will cause risk to the pipes and containers particularly at high temperatures. In this study, an in-house-developed vapor pressure test system was employed to measure the vapor pressure of MgCl₂–KCl binary salt. The schematic of this system is shown in Fig. 4. The entire system includes a sealed quartz tube to hold salt, a condenser used to condense most of the vapor salt to liquid and drop back to the bottom of the tube to prevent the pressure transducer from being clogged by the rising salt vapor, a furnace (Bartlett, Model 3K) employed to heat up the eutectic salt, a vacuum pump to make the pressure inside the quartz tube almost zero before test, a K-type thermocouple (Omega Engineering, Inc.) protected by a very fine quartz tube that is inserted into the molted salt to measure the temperature, a pressure transducer (PX419-100A5V by Omega Engineering, Inc., in accuracy of ±0.08% and maximum pressure of 6.89 × 10⁵ Pa) connected to the test tube to measure the pressure of samples, and a computer with LabVIEW data acquisition system used to collect data of temperatures and pressures. Wang et al. [21] had verified the reliability of this system in the previous work. The vapor pressure of ZnCl₂ single salt was measured and compared to the value from published literatures to calibrate this system. The difference of the data collected from this system and published equation was less than 20% at low temperature sections and less than 15% at high temperature sections.

The vapor pressure of the MgCl₂–KCl eutectic salt increases with the temperature, and is less than 13 kPa even at the temperature as high as 800 °C. This low vapor pressure also implies that the mass loss of MgCl₂–KCl molten salts in an open container is very small.

Viscosity is another important property affecting the pumping power and heat transfer of HTF in a CSP system. A high temperature viscometer (P610 by Theta Industrials, Inc.) was employed to measure this property for MgCl₂–KCl molten salt. The accuracy of this equipment was checked in our previous work by measuring the viscosity of a NaCl–KCl–ZnCl₂ salt sample in molar ratio of 20–20–60% at temperatures below 300 °C [21]. The results in that work [21] proved that this equipment was reliable and accurate since the difference of the measured data to literature-reported data is less than 1%.

An in-house developed density meter based on the Archimedean law was employed to conduct the experimental test for the density of MgCl₂–KCl eutectic molten salt. The calibration of this device was completed in previous work [15]. The density of pure NaCl and KCl were measured using this system at 1200 °C, and the difference is less than 1% when compared to data published in literature.

where $α,ρ,$ and $Cp$ are thermal diffusivity, density, and specific heat capacity, respectively.

The measured thermal diffusivity data were shown in Fig. 8. Under each temperature, there are ten measurements. The average, standard deviation, and measurement uncertainty at confidence interval of 95% are given in Table 8. It is seen that the thermal diffusivity has a rather small variation in between 0.27 and 0.29 mm²/s in the temperature range from melting point to 800 °C.

The calculated values for thermal conductivity $k¯$ and its measurement uncertainty $uk¯$ at several temperature points T are given in Table 9. The uncertainty of curve fitting for thermal conductivity is 6.75 × 10⁻³ W/m K. The overall uncertainties of the correlation (Eq. (9)) are obtained and given in Table 9.

Finally, all the thermophysical properties are summarized in Table 10. The provided uncertainties at a confidence interval of 95% for heat capacity, vapor pressure, viscosity, density, and thermal conductivity are the maximum of all the tested point. For the uncertainties of properties at a specific temperature, readers can refer to the corresponding figures in Secs. 3.1–3.5. The figure-of-merit analysis [22,23] about the thermophysical properties of molten salt MgCl₂–KCl and other molten salts such as NaCl–KCl–ZnCl₂ and NaCl–KCl–MgCl₂ have demonstrated that MgCl₂–KCl can serve as an excellent heat transfer fluid except its high melting point, which may be a concern.

Corrosion is a critical issue in the application of molten salt as HTF in a CSP system. Some investigations [24,25] about the mechanism of metal corrosion in molten chloride salts have pointed out that when the metal ions in the salts are less noble than the metals in the pipes and containers, they will not oxidize the more noble metals in the pipes and containers. Therefore, any metal corrosion should be attributed to the trace water and oxygen in the molten salts KCl–MgCl₂. With the molten salts being pretreated for elimination of water and oxygen, this study is interested in the tensile strength of the metals which have exposure to the molten salts at high temperatures. The tensile strength of the selected nickel-based alloys for pipes and containers was measured before and after a certain time of corrosion in the treated molten salt at a temperature of 800 °C.

A subsize specimen of a Haynes 230 and Hastelloy C-276 metal coupon in a “dog-bone” shape was used to test tensile properties in accord with ASTM E8, tensile testing of metallic materials [26]. The geometry and dimensions of the coupons are given in Fig. 10.

A test coupon was vacuum sealed into a quartz tube with Argon gas or with pretreated molten salt (KCl–MgCl₂ or NaCl–KCl–ZnCl₂ free of air and moisture) as shown in Fig. 11, and then is kept for a different length of time at a temperature of 800 °C.

The salt preparation/treatment and the sealing of coupons and salt in quartz tube are made through the following procedure. First, the molten salt (about 100 g) was heated to 500 °C and then kept at this temperature while sparging with Ar gas from a quartz capillary with 1 mm inner diameter at a flow rate of about 100 cm³/min for 1 h. Next, the coupon was immersed in the molten salt at 500 °C, while Ar gas was still flowing in the salt. The Ar gas was kept flowing into the salt for 5 min after metal sample immersion to remove any moisture adsorbed on the metal from air. Later, the Ar gas sparging tube was taken out of the salt, and then, the metal in salt was sequentially subjected to vacuum (−0.9 atm) for 5 min and Ar gas backfill for 1 min. Each evacuation and backfilling step was done five times over a total time of 30 min. Finally, the quartz tube containing salt and metal coupon was sealed shut while under vacuum (−0.9 atm) by melting the quartz tube with an oxyacetylene flame. During this process, the salt was kept in the molten state at 500 °C until transferred to a furnace. For corrosion test, the furnace accommodating the quartz tube with salt and coupon was kept at a temperature of 800 °C. After being heated at 800 °C in Argon and pretreated molten salt for certain period of time, the tensile strength curves of the coupon were obtained at room temperature and compared with that of a “cold” coupon (with no heating and salt exposure).

Figures 12 and 13 shows the test results of tensile strength versus strain for Haynes 230 (in Fig. 12) and Hastelloy C-276 (in Fig. 13) for a cold coupon and after exposure to a hot environment of air, Ar gas, and anaerobic molten salts (MgCl₂–KCl with molar ratio of 32–68% or NaCl–KCl–ZnCl₂ developed in our previous work [15] with mole ratios of 13.4–33.7–52.9%).

It is clear in Fig. 12 that a coupon with no exposure to high temperature environment showed the maximum strain at the ultimate strength, and the ultimate strength is also high. All the coupons either kept in Ar gas or in molten salts at 800 °C have unchanged slope of the strength-strain in the elastic region (below yield strength of 300 MPa) and the yield strengths also have no change. The strain at the ultimate strength has some difference due to the corrosion conditions. Under Ar gas condition given by curves 3, 4, 5, 14 corresponding to 1 week, 4 weeks, 8 weeks, and 16 weeks, the ultimate strength has some variation but does not show clear time-related decrease. Under molten salt KCl–MgCl₂ conditions, given by curves 6, 7, 8, 12, corresponding to 1 week, 4 weeks, 8 weeks, and 16 weeks, the ultimate strength does not show decrease compared to the coupons exposed to Ar gas. This is a clear indication that the corrosion of the coupons in molten salt KCl–MgCl₂ is very small, because metal exposed to molten salt does not show significant difference in the ultimate strength compared to coupons only exposed to Ar gas. Under molten salt NaCl–KCl–ZnCl₂ conditions, given by curves 9, 10, 11, 13, corresponding to 1 week, 4 weeks, 8 weeks, and 16 weeks, the maximum yield strength show a slight decrease compared to the coupons exposed to Ar gas. However, the ultimate strength of coupons does not show linear decrease with longer time of corrosion. Therefore, the corrosion of the coupons in molten salt NaCl–KCl–ZnCl₂ is also considered as rather minor.

In Fig. 13, the coupon with no exposure to high temperature environment again showed the maximum strain at the ultimate strength, and the ultimate strength is also high. The effects of Hastelloy C-276 exposures were similar to Haynes 230. At 800 °C, the coupons of Hastelloy C-276 either kept in Ar gas or in molten salts have unchanged slope of the strength-strain in the elastic region (below yield strength of 400 MPa), and the yield strengths also have no change. The strain at the ultimate strength is much smaller for the coupons exposed to high temperatures and corrosive environments. Under Ar gas condition given by curves 2, 3 corresponding to 1 week and 4 weeks, the ultimate strength does not show decrease, but the strain at the ultimate strength decreases with time. This indicates that even with no corrosion the long time under high temperature environment can still cause the ultimate strain decreases. Under molten salt KCl–MgCl₂ conditions, given by curves 4, 5, 8, 9, corresponding to 1 week, 4 weeks, 8 weeks, and 16 weeks, the ultimate strength does not show much difference compared to that of the coupons exposed to Ar gas. Again, this is a clear indication that the corrosion of the coupons in molten salt KCl–MgCl₂ is very minor, because metal coupons exposed to salt do not show difference in the ultimate strength compared to coupons only exposed in Ar gas. Under molten salt NaCl–KCl–ZnCl₂ conditions, given by curves 6, 7, 10, 11, corresponding to 1 week, 4 weeks, 8 weeks, and 16 weeks, the ultimate strength shows a slight decrease compared to that of the coupons exposed to Ar gas. However, the ultimate strength of coupons does not show linear decrease with longer time of corrosion, for example, between that of 4 weeks, 8 weeks, and 16 weeks. This is an indication that the corrosion of the coupons in molten salt NaCl–KCl–ZnCl₂ slows down significantly and does not increase linearly with time.

The thickness of corrosion layer on the surface of the dog bone samples exposed to molten salt was examined using scanning electron microscopy and energy dispersive spectroscopy (not shown in this paper). The thickness of the corrosion layer on the subsize specimens is on the orders tens of microns, and the increase rate of the thickness of the corrosion layer in time decreases with increasing time. The observed tensile strengths are consistent with the corrosion layer not affecting the underlying noncorroded metal and the metal corrosion decreasing as time goes on (from 1 to 16 weeks) when either metal (H230 and C276) is exposed to either anaerobic salt (KCl–MgCl₂ and NaCl–KCl–ZnCl₂) at 800 °C. Decreasing metal corrosion in time can be rationalized as due to the increasing depletion of residual oxidants (water and oxygen) from the anaerobic salt or/and the increasing growth of a barrier layer to the diffusion of corrosion-susceptible metals (like Cr) from the bulk to the surface of the metal as time goes on.

In this paper, the thermal and transport properties of MgCl₂–KCl binary eutectic molten salt with molar fraction of 32–68% were obtained from experimental tests. The tensile strength of specimen of metals by Haynes 230 and Hastelloy C-276 were measured before and after the samples were exposed to treated molten salts and high temperature of 800 °C. The conclusions from the investigation are as follows:

1. (1)

   The specific heat capacity has a very small variation in the liquid state, it is within the range of 0.990–1.013 J/g K.

2. (2)

   The measured average melting point of this molten salt is 424.4 °C with 1.3% relative error compared to the theoretical melting point. The average value of heat of fusion is 207.0 kJ/kg, with 95% confidence interval of uncertainty as 7.536 kJ/kg.

3. (3)

   The vapor pressure of current molten salt is less than 13 kPa even when the temperature is as high as 800 °C.

4. (4)

   The viscosity of this molten salt is below 4 cP when the temperature is higher than 600 °C. At the temperature around 430 °C, this molten salt has viscosity in the range of 6–7 cP.

5. (5)

   The density of this eutectic salt is in the range 1440–1730 kg/m³ and decreases with a (55.2 kg/m³)/100 °C slope against temperature increase.

6. (6)

   The thermal conductivity is obtained through the experimental data of heat capacity, density, and thermal diffusivity. It is in the range of 0.465–0.424 W/m K in the temperature interval of 450–800 °C and have minor decrease with the increasing of temperature.

7. (7)

   Hastelloy C-276 and Haynes 230 samples exposed to MgCl₂–KCl binary molten salt at 800 °C have almost no strength loss compared to those treated in Argon gas and NaCl–KCl–ZnCl₂ ternary molten salt no matter if the corrosion period is 1 week or longer. Haynes 230 alloy shows better corrosion resistance to MgCl₂–KCl molten salt than Hastelloy C-276.

Based on the experimental results above, MgCl₂–KCl eutectic molten salt with molar ratio of 32–68% is satisfactory and acceptable to serve as a high temperature heat transfer fluid in a CSP plant.

University of Arizona also provided some cost-share to this project. The authors also gratefully acknowledge that the tensile tests were performed as a service by Zachary Wing of Advanced Ceramics Manufacturing LLC in Tucson, AZ. The tensile data were plotted by Valeriia Bil, a Research Scientist working at the University of Arizona with D. Gervasio.

• U.S. Department of Energy (No. DE-EE0005942).