Molten Salt Spectroscopy for Quantification of Radiative Absorption in Novel Metal Chloride-Enhanced Thermal Storage Media Free

Molten inorganic salts hold a great deal of promise as high-temperature heat transfer fluids and thermal storage media in renewable energy applications, and they have found use in nuclear and solar thermal power [1,2]. In solar energy conversion, the most widespread use has been made of the nitrate salts—namely, the eutectic of potassium nitrate (KNO₃) and sodium nitrate (NaNO₃), commonly known as “solar salt.” This nitrate eutectic has been used in a variety of parabolic trough—and central receiver—type solar thermal facilities as the storage medium in a two-tank thermal energy storage configuration [2,3].

The next generation of solar thermal plants will likely need higher melting materials for their thermal storage systems. Receivers for use in the solar power tower arrangement have reached temperatures as high as 750, 812, and even 950 °C [4–6]. At these elevated temperatures, salts such as sodium chloride (NaCl), potassium chloride (KCl), and the eutectic (minimum melting) mixture of the two salts [(K–Na)Cl] present themselves as nearly ideal storage media candidates, owing to comparatively low cost, high heats of fusion, and melting temperatures in the range of interest (for NaCl, 800.7 °C, for KCl, 770.9 °C, and for the eutectic, 657 °C; see Table 1) [7–11]. It is likely that, as central receivers in the power tower arrangement achieve higher operating temperatures, materials such as these may be used as heat transfer fluids (e.g., in a dual tank storage system) or as PCMs. The latter provide the potential for increased energy storage density for limited ranges of operating temperatures owing to characteristically high heats of fusion, and moreover they are desirable due to their consistent energy delivery at design temperature [12].

These salts, however, are typically hampered by relatively low thermal conductivity (less than 1 W/m K in the molten state) [13]. To that end, many conductivity enhancement strategies have been explored. These include addition of greater conductivity materials, either as low concentration dispersed additives or in greater amounts as components of salt-based composites [14,15]. Much work has also been done in the design of the containment—e.g., use of heat pipes or curved slabs to improve conduction and/or convection [16,17].

This study explores a novel heat transfer enhancement strategy for the storage material itself. At the higher temperatures, where these salts would be applied as storage media (>500 °C), radiation becomes a significant mode of heat transfer. It is proposed that inclusion of small amounts of IR-absorptive additives in highly IR-transmissive salts (e.g., alkali metal halides) will allow for greater radiative heat transfer, thereby compensating for characteristically low thermal conductivity.

Determining the degree of improvement requires measurement of the absorption coefficient in the spectral range of interest. Unfortunately, traditional spectroscopic methods are ill-suited for work with molten salts: Direct transmittance measurements overestimate absorption if they do not capture backscattering, and many window materials are prone to corrosive attack by these molten salts [18]. Another strategy to employ would be attenuated total reflectance spectroscopy [19]; but, here again, owing to the close contact between the reflecting prism and liquid, material compatibility renders this method impractical. The authors therefore developed a new IR-reflectance apparatus (Fig. 1) to allow for determination of the spectral absorption coefficient of the newly formulated storage media in the molten state. The apparatus consisted of an alumina crucible coated alternately at the bottom with a reflective (platinum) or absorptive (graphite) surface, a ceramic heating element housing the crucible, and a combination of zinc sulfide (ZnS) and zinc selenide (ZnSe) windows for containment of the salt and allowance of inert purge gas flow, respectively.

This paper specifically addresses the eutectic mixture of potassium and sodium chloride, (K–Na)Cl. Both KCl and NaCl are highly transmissive in the spectral region of 2.0–13.0 micrometers (μm), and it is worth noting that at the melting point of the eutectic (657 °C), nearly 90% of spectral blackbody emissive power falls within this region, as given by the Planck distribution [20]. Various transition metal chlorides of known solubility in KCl and/or NaCl were investigated as potential absorptive additives, including cuprous chloride (CuCl), cobaltous chloride (CoCl₂), ferrous chloride (FeCl₂), and nickelous chloride (NiCl₂). Using the reflectance apparatus, IR spectra were obtained for these additives at less than 0.2 weight percent (wt.%) concentration in (K–Na)Cl and improved IR absorption, and radiative transfer properties were demonstrated. Using these spectra, estimations of the absorption coefficient of the additives in the pure salt were calculated. Further, thermophysical properties relevant to thermal energy storage (i.e., melting temperature and latent heat of fusion) were measured via differential thermal analysis (DTA) for the pure and additive-enhanced PCMs.

A general treatment of the absorptive properties of inorganic salts is given in Modest's description of the Lorentz model, shown to adequately capture the absorption behavior of ionic crystals [21]. This model predicts that such materials should have one or more absorption bands in the mid- to far-IR (Reststrahlen absorption) and several absorption bands near the ultraviolet (electronic excitation)—between these absorptive (opaque) regions, ionic crystals are generally transparent. A discussion of the quantum-mechanical considerations behind absorption and emission of photons in these systems is beyond the scope of this paper; for detailed discussions of this topic the reader is referred elsewhere [22–24].

In speaking of the absorptive properties of these materials and those of KCl and NaCl in particular, it is worthwhile to note the expansive reference work of Touloukian and DeWitt, which collects the radiative properties (reflectance, transmittance, emittance, etc.) of various nonmetallic solids [25]. According to data compiled in this reference, KCl is highly transmissive (transmittance greater than 90%) in the spectral region of 0.4–15.0 μm, for crystals 10–12 mm thick; analogously, NaCl is highly transmissive in the region of 0.4–13.0 μm, for crystals 2.8–6.5 mm thick. Other halides are similarly transmissive over much of the IR including calcium fluoride, which has also been proposed for thermal energy storage applications [26]. It bears mentioning that these measurements were made at or near room temperature; the near- to mid-IR absorption of such crystalline solids, especially if appreciable concentrations of lattice defects or dopants are present within the crystal, can be expected to rise significantly with temperature [21]. (Further information on the radiative properties of materials can be found in other references [27,28], as well as in various textbooks [20,21,29].)

Of course, the high transmissivity of either KCl or NaCl does not guarantee that the mixture of the two will be itself transmissive; indeed, at room temperature, one will notice that the solidified (K–Na)Cl salt is quite opaque to visible radiation, as the cooling solid will self-diffuse to form closely packed lamellae of alternatingly high concentration KCl and high concentration NaCl phases. Importantly, though, the salt regains its high transmissivity as it heated to elevated temperatures with a consolute point at approximately 500 °C [30]. At this point, the salt becomes a homogeneous solid solution which is transparent to visible radiation.

The (K–Na)Cl eutectic has been studied extensively, and a review of pertinent literature yields the properties data relevant to our study here. A collection of these data is compiled in Table 1 with properties of the pure component salts included for reference. The first entry of the table for the eutectic salt was obtained from the thermophysical properties software factsage [8]; the second entry presents calculated values obtained from analytical formulations of the system's thermodynamic mixing properties [9]. The remaining entries are experimental data. It can be observed there is significant variability in the data. It is hoped that thermophysical properties data collected for this current study will serve to clarify the validity of some of these values. For the purposes of this study, the molar composition of the eutectic was taken as 50.02:49.98 KCl:NaCl and the density of the molten salt was taken as 1.58 g/ml. More detailed discussion of the various studies performed and their assertions concerning the composition and melting point of the (K–Na)Cl eutectic salt can be found in the referenced literature [9,30].

The additives for potential radiative absorption enhancement have been chosen on the basis of two criteria:

1. (1)

   Potential for IR absorption

2. (2)

   Solubility in the molten salt

With regard to the first criterion, previous work has shown the potential for transition metal oxides—namely, Co₃O₄ and CuO—to increase radiative absorption in the ultraviolet, visible, and near-IR regions of the electromagnetic spectrum [31]. Further, studies into the middle and far-IR show broadband absorption for such transition metal oxides [32]. While these materials are no doubt promising for IR absorption, there is an attendant complication to the use of insoluble additives: maintenance of their suspension in the salt. For PCMs especially, which would likely be sequestered in a containment of some kind (e.g., capsule), continuous agitation may be difficult, and such additives may aggregate and settle.

As such, the second criterion, molten salt solubility, was examined critically during the selection process. It was determined that the four additives ultimately selected—CuCl, CoCl₂, FeCl₂, and NiCl₂—were each miscible in KCl and NaCl as binary solutions; indeed, in the case of FeCl₂, there is strong suggestion that it would form a solid state compound with KCl (double salt), K₂FeCl₄, which would allow for improved distribution in both the solid and liquid states [33].

Here, $δ$ is the film thickness and $θi$ is the angle of incidence of the radiation. For the situation under consideration here, we can reasonably specify the thickness of the layer (using a specific mass and known density) and the angle of incidence (set by the geometry of the reflectance apparatus). Thus, for any single reflectance measurement, there are three unknown quantities:

1. (1)

   The reflectivity at the surface of the salt (i.e., nitrogen–salt interface), $ρ1$⁠.

2. (2)

   The reflectivity at the salt–substrate interface, $ρ2$⁠.

3. (3)

   The absorption coefficient, $κ$⁠.

In our case, the reflectivity of the salt–substrate interface is measured separately: The reflective substrate (platinum) is measured relative to a silvered mirror. The absorptive substrate (graphite) is taken to have reflectivity of zero. Hence, from measurement of the reflectance of the salt over graphite, we obtain $ρ1$⁠; given this value and the known value of $ρ2$⁠, measurement of the reflectance of the salt over the platinum value yields the transmissivity of the salt layer. Given the transmissivity of the pure salt and that of the additive–salt system, the absorption coefficient, $κ$⁠, of the additive in the pure salt can be calculated directly.

All IR spectra were obtained using a Jasco FTIR-6300 Fourier-transform infrared spectrometer. In this system, a high-intensity ceramic source was used in conjunction with a mercury cadmium telluride detector, effectively limiting the observable spectral range to 7800–650 cm⁻¹ (although the range into the far-IR is further limited by the window materials and the salts themselves).

The heated sample chamber was provided by Pike Technologies of Madison, WI. Designed for diffuse reflectance applications, the chamber consists of a heated ceramic sample holder to be used in conjunction with porous alumina crucibles. The heater is capable of reaching 900 °C, and the chamber is cooled by chilled water pumped at a rate of approximately 50 ml/min. The cooling water keeps most of the sample chamber at approximately room temperature, so that, for instance, during emission measurements, background radiation interference is minimized. Using a suitable window (in this case, ZnSe), inert gas purge can be passed through the chamber, with the pores of the alumina crucibles allowing the purge gas to pass directly through the (diffuse) powdered sample.

For work with regular reflectance of molten salts, some additional features were introduced to the sample holder. First, porous alumina pans would not allow for liquid samples; therefore, solid alumina pans were used. However, these pans provide a hindrance to the flow of the inert purge gas. The pans were therefore scored across the bottom in a cross-hatch manner. These grooves allowed purge gas to flow around the sample and beneath the pan to the purge outlet.

Important study of the reflectance and optical properties of molten nitrates [36], chlorides [37], and carbonates [38] was undertaken by Makino et al., and a thorough description of their apparatus has been given in Ref. [36]. The basic strategy of this paper's measurement of radiative properties is similar: Regular reflection is measured of molten salt over a reflecting background. However, our study adds the step of performing a separate measurement over black background to elucidate the absorption coefficient.

It was found that for high vapor pressure salts (notably, metal halides), evaporated salt would condense upon and outermost ZnSe window and severely degrade it. Therefore, a second additional feature was introduced: A salt containment window to be placed directly over the sample crucible such that significant amounts of salt vapor would not be allowed to escape. A variety of materials were tested for this purpose; their properties have been presented elsewhere, and these data are presented in Table 2 [39]. KCl, NaCl, potassium bromide (KBr), and calcium fluoride (CaF₂) are all promising as window materials, but they are prone to ion exchange with the evolved salt vapors tested here. Hence, they would only serve for one use. (It is possible that these materials could be used as pressed powders, significantly reducing the cost of the window material; the authors are currently exploring this option.) Also, ZnSe, too, showed a high susceptibility to attack by the salts and also by the high temperatures alone. ZnS alone showed the desired resistance to degradation upon multiple uses. While the transmissive range of ZnS crystals is smaller than some of the other materials, it should cover the spectral range relevant here—namely, 2.0–13.0 μm, in which 90% of blackbody emissive power at 657 °C is captured. (CaF₂, on the other hand, would cut off this range on longer wavelength side.) A schematic of the dual-window reflectance cell is shown in Fig. 1.

It should be emphasized that the relation above is accurate for spectral quantities—e.g., owing to spectral variability of radiative properties, the same relation does not apply to total reflectivity and emissivity. Also, strictly speaking, this yields the normal incidence spectral reflectivity; due to the relatively small angles of incidence considered here, no correction shall be applied for any directional variation in the following analyses. The blackbody reference used was an IR-563/301 manufactured by Infrared Systems Development Corporation of Winter Park, FL.

DTA measurements were performed using a TA Instruments SDT-Q600 TGA/DSC. A three-point calibration of the instrument was performed using high purity tin, zinc, and sodium fluoride as calibrants. The cell constant (latent heat) was calibrated using high purity zinc. Samples were placed in platinum pans to prevent the creeping of salt samples during melting and solidification. In each case, samples were melted twice; the second run (after formation of the binary/ternary system) was used to obtain thermophysical properties. Multiple samples for each material were tested, and averaged values were obtained. All measurements were made under argon purge flow at a heating rate of 10 °C/min.

A novel method was applied to prepare samples for study. Given the potential for the additives to solid solutions or double salts with bulk NaCl and/or KCl, the following strategy was employed:

1. (1)

   Additives were mixed at a concentration of 0.4 wt.% in agate mortar with spectroscopic-grade NaCl in a glovebox under inert gas.

2. (2)

   The mixtures were heated to approximately 250 °C for 24 hr in the glovebox to encourage formation of solid solutions.

3. (3)

   The heated mixtures were then mixed in agate mortar with the required quantity of KCl (99.997% purity, metals basis) [8]. The resultant mixtures were approximately 0.18 wt.% additive.

4. (4)

   Throughout the following analyses they were kept in a vacuum desiccator or drying oven kept at 110 °C.

These moisture control steps were important, since three of the four additives considered (excepting CuCl) are known to be very hygroscopic. The concentrations of the salt systems studied are shown in Table 3.

The measured reflectance of the platinum substrate is shown in Fig. 2. The reflectivity is comparatively high, and so will serve the purpose of providing a reflective background to the semitransparent molten salts.

Next, the salts' spectra were measured. Upon heating, it was observed that the characteristic spectra of the loose powder samples changed dramatically upon melting, thus making it a simple matter to determine when to collect the spectra. That said, the general procedure involved heating the sample/background to approximately 680 °C, which ensures the melting would take place rapidly. Spectra were collected as soon as the melt had occurred, which could be verified by visual examination of the sample.

In Fig. 3(a), the spectra obtained for the pure eutectic salt are presented—namely, the reflectance of the overall system (⁠$R$⁠), the reflectivity of the salt–nitrogen interface (⁠$ρ1$⁠), and the calculated transmissivity (⁠$τ$⁠) from the net radiation method [34]. (The reflectivity of the platinum substrate, $ρ2$⁠, is illustrated in Fig. 2.) Similar spectra are obtained for each of the salt–additive systems, and these are depicted in Fig. 3(b).

Given the transmissivity of the salt systems, it is possible to make an estimation of the absorption coefficient of the additive materials in the pure eutectic salt. For these calculations, the density of the salt is assumed unchanged upon addition of the small concentration (∼0.18 wt.%) of additive—i.e., the film thickness, $δ$⁠, is assumed to be constant in each case. The angle of incidence for the geometry considered here is approximately 30.6 deg. The calculated absorption coefficients for each additive–salt system are presented in Fig. 4. Note that these measurements show absorption relative to the pure fused eutectic salt (e.g., see Fig. 3(a)), which itself was found to have a transmissivity of approximately 0.9.

The spectra obtained agree qualitatively with expected behavior: The pure eutectic appears mostly transparent. Inclusion of the additives results in a marked decrease in the measured reflectance, especially in the case of FeCl₂. From the calculated absorption coefficients, it is clear that the additives increase IR absorption, and therefore they are promising candidates for heat transfer enhancing additives. One exception is the case of NiCl₂ at wavelengths greater than 12 μm. Here, the absorption coefficient is approximately zero, and it is likely that the additive has no effect on the absorption of the pure eutectic in this region. It is also prudent to note here that the FeCl₂ measurement was complicated by the formations of oxides, which likely exaggerated the degree of attenuation in this case. This was the only sample that displayed any evidence of oxide formation—it is possible that the other additives are less prone to this issue, although it may simply have been an error of the experimental apparatus (e.g., greater infiltration of oxygen or moisture into the reflection cell). Further measurements are planned, which will take care to void the sample chamber of oxygen/moisture prior to measurement. In any event, this behavior deserves consideration if these materials are to be employed in larger-scale thermal energy storage units. Oxidation of the material can be avoided, for instance, through atmospheric control, use of a preferential oxidizer (i.e., getter), etc.

The thermophysical properties measured by DTA are presented in Table 4. As can be seen, the heat of fusion and melting point for the pure eutectic system agree well with literature values [9]. Note that while the melting temperatures of the additive-enhanced materials are slightly lower than that of the pure salt (as one would expect by inclusion of a solute), the latent heat of fusion, of primary importance for PCM storage media, seems little impacted. Indeed, in the case of the divalent additives especially, there is an apparent increase in latent heat, although this increase lies well within the error of measurement. It is not unreasonable that latent heat would increase upon addition of small amounts of soluble additive, and this phenomenon is observed in the NaCl–KCl system. Addition of small amounts of KCl to pure NaCl increases the latent heat of fusion, most likely because the presence of the larger impurity potassium ions creates regions of stress that attract and “pin” the moving lattice vacancies required for phase change (i.e., more energy required for melting) [10]. Of course, at higher concentrations, the competing effect of weakening the lattice causes the latent heat to decrease (i.e., less energy required for melting). From the available data, though, it seems that the latent heat is approximately unchanged upon addition of the additive. The melting point, also, is little changed by addition of additives, with the highest melting point depression, approximately 1.0 °C, occurring in the CuCl–(K–Na)Cl system. In either case, the additive does not significantly impact the performance of the PCM as a thermal energy storage medium.

An IR-reflectance apparatus was developed for testing the radiative absorption of novel inorganic salt thermal storage media in the molten state. The method of collection of the overall layer-substrate reflectance and the salt surface reflectivity by means of two different substrate materials (platinum and graphite) has been demonstrated. The inclusion of transition metal chloride additives shows distinct increases in absorption in the IR, with only slight impacts on melting temperature and latent heat of fusion.

Further work currently underway includes the collection of spectra from additional samples to determine the error in measurement method, as well as from different concentrations to estimate the concentration dependence of radiative absorption. Part of this study includes working with thinner windows made of pressed halide powders; as mentioned above, these materials would be good for only one use, but they may prove a good choice due to their significantly lower cost, relative to polished crystal optics. It would also be prudent to test the results obtained here with a method based on the emissivity of the substrate/salt system, as has been described by Makino [40]. Such efforts are currently underway. Also, the specific heat capacity of these systems is being measured in the solid and, if possible, molten states. It is not anticipated, however, that such small concentrations of additives will have a significant impact on this property of the salt.

Radiative transfer in thermal energy storage systems is valuable area of research, in light of the fact that radiation can play a major role in heat transfer in the next generation of high-temperature thermal energy storage. Spectroscopy, whether through transmission, reflection, or emission measurements, is a direct means of characterizing the radiative properties of materials needed in these systems. It is hoped that this paper will encourage further work in the field of radiative properties measurement in general and molten salt spectroscopy in particular.

The authors are very grateful for the assistance and advice provided by colleagues and staff at the Clean Energy Research Center at the University of South Florida. The work presented herein has been funded in part by the U.S. Department of Energy (DOE) through the Advanced Research Projects Agency-Energy (ARPA-E) under the Award No. DE-AR0000179.

$Cp$ =

specific heat capacity

(K–Na)Cl =

potassium–sodium chloride eutectic salt

$R$ =

overall reflectance

$T$ =

temperature

$Tm$ =

Reynolds number and similar abbreviations do not use italics

$δ$ =

layer thickness

$ΔHfus$ =

latent heat of fusion

$ɛλ$ =

spectral emissivity

$θi$ =

angle of incidence

$κ$ =

absorption coefficient

$ρλ$ =

spectral reflectivity

$ρ1$ =

reflectivity of first interface

$ρ2$ =

reflectivity of second interface

$τ$ =

transmissivity of the layer