DEVELOPMENT OF MOLTEN SALT HEAT TRANSFER FLUID
WITH LOW MELTING POINT AND HIGH THERMAL STABILITY
Justin W. Raade1 and David Padowitz
Halotechnics, Inc. 5980 Horton St. Suite 450, Emeryville, California 94608 USA. +1 (510) 547-2634
1 Corresponding author: firstname.lastname@example.org
This paper describes an advanced heat transfer fluid (HTF) consisting of a novel mixture of inorganic salts
with a low melting point and high thermal stability. These properties produce a broad operating range molten
salt and enable effective thermal storage for parabolic trough concentrating solar power plants. Previous
commercially available molten salt heat transfer fluids have a high melting point, typically 140 °C or higher,
which limits their commercial use due to the risk of freezing. The advanced HTF exploits eutectic behavior
with a novel composition of materials, resulting in a low melting point of 65 °C and a thermal stability limit
over 500 °C. The advanced HTF described in this work was developed using advanced experiment design
and data analysis methods combined with a powerful high throughput experimental workflow. Over 5000
unique mixtures of inorganic salt were tested during the development process. Additional work is ongoing to
fully characterize the relevant thermophysical properties of the HTF and to assess its long term performance
in realistic operating conditions for concentrating solar power applications or other high temperature
Keywords: molten salt, eutectic mixture, heat transfer fluid, thermal storage.
This paper describes an advanced heat transfer fluid (HTF) consisting of a mixture of inorganic salts for use
in concentrating solar power (CSP) applications or other high temperature processes. Previous commercially
available molten salt heat transfer fluids have a high melting point, typically 140 °C or higher. The advanced
HTF described in this work exploits eutectic behavior with a novel composition of materials, resulting in a
low melting point of 65 °C and a thermal stability limit over 500 °C.
Concentrating solar power uses mirrors to focus solar energy to boil water and make high pressure steam.
The steam subsequently drives a turbine and generator unit to generate electricity. There is a need to bring
CSP electricity cost down to the point of being competitive with traditional fossil fuel-based electricity. An
advanced low melting point HTF with a high thermal stability is a key technical advance necessary to reduce
the cost of CSP electricity. This novel material would enable higher temperature operation and increased
efficiency in converting solar energy to electricity. Increasing the maximum fluid output temperature of
current CSP plants from 390 °C to 500 °C would increase the conversion efficiency of the Rankine power
block, thereby reducing the levelized energy cost by 2 cents/kWh. Achieving 500 °C operation would also
double the effectiveness of sensible heat thermal storage systems, significantly reducing the capital cost of
thermal storage .
The state of the art HTF used in CSP applications is an organic substance that is a eutectic mixture of
biphenyl and diphenyl oxide, sold under the brand names VP-1 and Dowtherm A , . It exhibits a low
melting point of 12 °C but is limited to an upper temperature of 390 °C due to thermal degradation. This
temperature ceiling limits the thermodynamic efficiency of the Rankine cycle steam turbines driven by the
heat from the solar collector field. The organic HTF exhibits a high vapor pressure at elevated temperatures,
approximately 10 bar at 390 °C, which prevents its use as a thermal energy storage medium. The cost of
commercially available organic HTF is quite high.
Molten salts exhibit many desirable heat transfer qualities at high temperatures. They have high density, high
heat capacity, high thermal stability, and very low vapor pressure even at elevated temperatures. Their
viscosity is low enough for sufficient pumpability at high temperatures, and many are compatible with
common stainless steels. Salts of many varieties are currently available in large commercial quantities from
several suppliers. There are several commercially available salt formulations, mixtures of nitrates or nitrites
sold under the brand name Hitec , . A drawback of these molten salts as heat transfer fluids however is
their relatively high melting point, typically 140-240 °C. The operational risk of a freeze-up if the process
temperature drops unexpectedly adds cost to systems that use salt as a heat transfer fluid. Additional
hardware must be installed, such as heat tracing, insulation, or emergency water-dilution systems . This
high melting point limits the practicality of molten salts as heat transfer fluids in CSP applications. The
advanced HTF described in this paper exhibits both a low melting point and a high thermal stability, as
shown in Figure 1. Table 1 shows the relevant properties of currently available high temperature heat transfer
fluids relevant for CSP applications.
° C ° C
Figure 1: Operating range of revelant CSP heat transfer fluids.
Name Manufacturer Components
VP-1 or Dowtherm A
Solutia or Dow
12 °C 400 °C
Hitec XL Coastal Chemical
140 °C 500 °C
Hitec Coastal Chemical
142 °C 538 °C
Hitec Solar Salt Coastal Chemical
240 °C 593 °C
Table 1: Currently available heat transfer fluids for CSP applications.
1.2. Eutectic mixtures
A eutectic mixture exhibits the lowest melting point of any similar mixture with the same components. The
change in Gibbs free energy ∆G of a substance at the melting temperature T can be expressed in terms of the
change in enthalpy ∆H and the change in entropy ∆S.
At equilibrium, ∆G = 0 and the melting temperature can be expressed as
Eutectic mixtures tend to disrupt intermolecular forces (reducing the change in enthalpy) or to increase the
disorder generated upon melting (increasing the change in entropy). This leads to a reduction of the melting
Eutectic behavior is common with binary mixtures of salts, and can be even more dramatic with ternary
mixtures. There has been significant work done both on modeling the phase behavior of binary and ternary
mixtures of salts, as well as experimentally measuring their behavior , . An example of a simple binary
salt mixture is sodium nitrate and potassium nitrate (NaNO3 and KNO3). Sodium nitrate melts at 307 °C and
potassium nitrate melts at 337 °C. This mixture has a eutectic point at 46 mol % NaNO3 and 54 mol % KNO3
which exhibits a drastically reduced melting point of 222 °C. This represents a melting point suppression of
85 °C from the lowest melting single component. An example of a ternary salt mixture is sodium nitrate,
potassium nitrate, and lithium nitrate (LiNO3). Lithium nitrate melts at 253 °C. The ternary mixture has a
eutectic point at 18 mol % NaNO3, 44.5 mol % KNO3, and 37.5 mol % LiNO3 which exhibits a melting point
of 120 °C. The addition of lithium nitrate to the mixture achieves an additional melting point reduction of
102 °C as compared to the binary mixture.
Eutectic behavior and more drastic melting point reduction occurs with more complex salt mixtures, such as
quaternary or higher order mixtures. There is limited experimental data of higher order mixtures; some recent
work has been done on quaternary mixtures of nitrate salts (with Li, Na, K, and Ca cations) with melting
points below 100 °C .
It is difficult to accurately model phase behavior of higher order salt mixtures. Detailed material properties of
each component must be known, some of which must be measured experimentally. Existing databases of
thermodynamic salt properties are incomplete, with many salts of interest (such as nitrates and nitrites)
missing. It is therefore typically more straightforward to rely on experimental methods and to directly
measure the phase transitions of a system of salts. The large number of possible combinations with higher
order mixtures makes experimental work burdensome, since the number of possible mixtures increases
exponentially with the number of components. Eutectic behavior is quite sensitive to the weight percent of
each component in the mixture; a deviation of only a few percentage points may have a significant effect on
the resulting melting point. If one assumes that each component in a salt mixture can be controlled to the
nearest percentage point, then with a two component system there are 101 possible combinations, with a
three component system there are 5151 combinations, and with a four component system there are 176,851
combinations. The permutations increase correspondingly when one varies the components to explore
different salt systems.
2. Experimental methods
2.1. Materials discovery workflow
In order to overcome the challenge of the large number of possible combinations of salt mixtures, the
advanced HTF was developing using high throughput experimentation techniques and apparatus. A materials
discovery workflow capable of screening up to 100 mixtures per day was developed, shown in Figure 2. Salt
mixtures were formulated using automated robotic systems for both powder dispense and liquid dispense.
The powder dispense system was the MTM Powdernium from Symyx Technologies (Sunnyvale, California).
This device measures each component as it is being dispensed and records the final weight with high
accuracy. It can dispense many different components to many different mixtures. The liquid dispense system
was the Synthesis Station Core Module from Symyx. The melting point of each mixture was measured with
the Parallel Melting Point Workstation (PMP) from Symyx. The PMP allows the melting point for each
mixture in the 96 well plate to be measured simultaneously. The PMP heats the plate at a controlled rate and
uses an optical method to record the temperature at which each mixture transitions from opaque to clear. This
transition corresponds to the liquidus temperature, which is defined as the temperature during heating at
which the last remaining solid phase melts and becomes liquid. The liquidus temperature is also equivalent to
the temperature during cooling at which a solid phase first appears in the melt (assuming no supercooling).
However, supercooling is common with molten salts and therefore only data acquired during a heating mode
was used to obtain the melting point. The melting point in this paper is defined as the liquidus temperature.
Mixtures that exhibited a low melting point were subjected to further testing for thermal stability using a
thermogravimetric analysis (TGA) device, the Q500 TGA from TA Instruments
(New Castle, Delaware). A TGA heats a sample in a controlled environment and
continuously measures the sample weight, which typically decreases at higher
temperatures as the sample decomposes into gaseous products.
2.2. Experimental procedure
A standard procedure was developed to prepare and characterize salt mixtures.
The first step was to prepare free flowing anhydrous salt components.
Components were purchased in reagent grade purity, typically 99% pure, from
Sigma Aldrich (St. Louis, Missouri). Each component that was available in
anhydrous form was ground with a mortar and pestle and dehydrated in an oven at
115 °C for at least 12 hours. Calcium nitrate was procured in a tetrahydrate form
and was not dehydrated prior to dispensing. Components were typically dispensed
as powder, but potassium nitrite was dispensed as an aqueous solution because it
is very hygroscopic and tends to form clumps when in powder form. Calcium
nitrite was only available from suppliers as an aqueous solution and was
dispensed in this form. The mixtures were dispensed into a borosilicate glass plate
containing 96 wells in an 8 by 12 array. Each mixture had a total mass of 250 mg.
After dispensing, the plate was placed in a furnace purged with nitrogen gas and
heated to 400 °C for at least 8 hours in order to ensure complete melting and
homogenization of each mixture. Mixtures that contained hydrated components or
water were first held at 150 °C for at least 4 hours to boil off the water without
spitting. After melting the plate was allowed to cool and stored in a dessicator
before subsequent testing. The plate was then inserted into the PMP and the
temperature was set to 50 °C and allowed to stabilize for 30 minutes. The
temperature was then ramped to 200 °C at 20 °C/hour.
Mixtures that exhibited a low melting point were subjected to further testing for thermal stability.
Approximately 20 mg of each mixture was scraped from its well in the glass plate and loaded onto a platinum
pan for TGA testing. The maximum temperature or thermal stability of a sample, termed ‘T3’, was defined
for screening purposes as the temperature at which it has lost 3% of its anhydrous weight during a TGA test
ramping at 10 °C/min. The anhydrous weight of a salt sample was defined as the weight at 300 °C during the
TGA test. Initial weight loss below 300 °C is due to absorbed water evaporating from the sample. Each
mixture was tested in two atmospheres, one of air and one of nitrogen, in order to observe the effect of
oxidation. The thermal stability using the T3 method typically produces similar results for each mixture in a
given system; however significant differences are observed between systems. Therefore only a representative
set of mixtures from each system were tested for thermal stability rather than every mixture in the system.
Mixtures with only nitrate typically have similar T3 values for air and nitrogen atmospheres. Mixtures
containing nitrate/nitrite typically have a higher T3 value when tested in air. This is likely due to oxidation of
the nitrite ion to nitrate, producing an increase in weight and masking the effect of weight loss due to thermal
decomposition. For this reason the T3 value in nitrogen is considered more useful for nitrate/nitrite mixtures.
2.3. Experiment design methodology
A rigorous experiment design methodology for high dimensional salt mixtures was devised. Over 5000
unique mixtures of inorganic salts were tested during the screening process. The literature for available
systems was first reviewed in order to provide a starting point for exploring uncharted phase space. Extensive
data exists for binary and ternary phase diagrams of inorganic salts . Data from ternary mixtures served
as a tool to predict which salts would result in the greatest melting point reduction. However, there is
minimal data available for quaternary (four component), quinary (five component), and higher order systems.
Custom software tools were developed in order to rapidly design constrained sets of experiments with the
desired salt components. These designs were fed into the high throughput materials discovery workflow. In
this manner many feasible salt types (nitrates, nitrites, carbonates, sulfates, and others) were surveyed in
Figure 2: Materials
mixtures containing up to six components.
2.4. Plotting experimental results with high order phase diagrams
The methodology and software tools for graphing six dimensional phase data were developed. The data can
include five degrees of freedom in composition (typically six salts or seven ions) plus one for the melting
point (depicted by color). The phase diagram is a graphical device that allows the composition and melting
point of mixtures to be represented simultaneously (this type of phase diagram is called a polythermal
projection). The typical phase diagram is triangular, which allows the plotting of a ternary system of three
salts (typically four ions). Each corner of the triangle represents a pure salt and the interior area represents
mixtures of varying proportions. The color represents the melting point. A quaternary system of four salts
(typically five ions) may be plotted by a series of triangular phase diagrams. The location of each ternary
diagram along a horizontal axis represents the proportion of the 5th ion. A quinary system of five salts
(typically six ions) may be plotted by a two dimensional surface of ternary phase diagrams (Phase Diagrams
for Ceramists, vol. 1 discusses this technique in the introductory material, for up to six ions). Each ternary
phase diagram is located at the (x, y) coordinates corresponding to the level of the 5th and 6th ions (ion 5, ion
6). A system of six salts (typically seven ions) may be plotted by a series of two dimensional surfaces of
ternary phase diagrams. Each surface represents a constant value of the 7th ion. Figure 3 shows examples of
these high dimensional phase diagrams.
Ion 5Ion 5
ion 6ion 6
Ion 5Ion 5
ion 5ion 5
ion 7ion 7
Figure 3: Graphical representation of six and seven ion phase space.
3. Results and discussion
3.1. Melting point
The advanced HTF developed as a result of the materials discovery screening process represents the lowest
melting point known for a nitrate-only salt mixture that is stable to at least 500 °C. Its properties are
described below, including the composition, melting point, and thermal stability. Laboratory scale quantities
of salt mixtures were tested during the course of this work, however a salt mixture of any desired size with
the same properties (melting point and thermal stability) can be prepared by increasing the amount of each
component but maintaining the relative proportions. The data shown here was collected from a ~250 mg salt
mixture: 18.6 mg of LiNO3, 14.4 mg of NaNO3, 53.4 mg of KNO3, 102.7 mg of CsNO3 and 62.1 mg of
Ca(NO3)2-4H2O. Table 2 summarizes this data. A U.S. provisional patent application has been filed (No.
61/325,725) that covers the composition of the advanced HTF.
Property Value Method Notes
Melting point 65 °C Experimental (PMP) Temperature is liquidus
Thermal stability limit
561 °C Experimental (TGA, 3%
weight loss at 10 °C/min)
To verify with field
Thermal stability limit
563 °C Experimental (TGA, 3%
weight loss at 10 °C/min)
To verify with field
Table 2: Selected physical properties of the advanced HTF.
Figure 4 shows the composition of the eutectic as well as a phase diagram of the mixtures in the immediate
neighborhood of the eutectic. This data represents the best known eutectic (lowest melting point) mixture of
the given salts, discovered after a thorough search of feasible salt mixtures. Figure 5 shows an expanded
phase diagram of the area further from the eutectic composition. This diagram shows four dimensions of
composition plus one more dimension for liquidus temperature.
KNO3 23% 30%
CsNO3 44% 30%
Ca(NO3)2 19% 15%
Figure 4: Composition of the advanced HTF (patent pending) and phase diagram showing immediate
neighborhood around eutectic with the liquidus temperature in degrees celsius.
Figure 5: Phase diagram showing expanded region around the eutectic composition with the liquidus
temperature in degrees celsius.
3.2. Thermal stability
The thermal stability using the T3 method is shown in Figure 6 for a mixture similar in composition to that
described in Figure 4. The large weight loss below 200 °C is due to the evaporation of absorbed water. This
data represents a laboratory screening measurement and may not provide a definitive prediction of the HTF
long-term thermal stability in a commercial application. The T3 method ranks the salt mixtures in order of
relative stability rather than acting as an absolute measurement of stability. It is primarily a laboratory-scale
screening test that gives a comparative ranking of candidate salt mixtures. Field testing is planned with
Sandia National Laboratories and industry partners to verify long-term performance.
Figure 6: Thermal stability behavior of Li-Na-K-Cs-Ca-NO3 mixture in air and nitrogen.
3.3. Cost of components
Cost is important consideration for a commercially viable heat transfer fluid. The components of the
advanced HTF described in this paper are expected to have the following relative cost (in order from least
expensive to most expensive): Ca(NO3)2 ? NaNO3 ? KNO3 ? LiNO3 ? CsNO3 . The per unit cost of a
single component is generally reduced when produced in larger quantities and at lower purity levels. The cost
of raw materials for this advanced HTF is likely to be considerably higher than the simple binary solar salt
(sodium and potassium nitrate). The cost may be reduced by optimizing the mixture to limit the amount of
cesium and lithium while still maintaining acceptable physical properties. Such properties in addition to the
melting point are important for a commercially viable heat transfer fluid, including the viscosity (affects
parasitic pumping losses), the long-term thermal cycling behavior (the composition may drift after many
melt/freeze cycles), and chemical compatibility with common steels (corrosion tendency). The optimal
combination of these properties at an acceptable cost may result in a composition differing from that given in
Figure 4. This optimization is currently underway at Halotechnics and its partners.
5. Conclusion and future work
This paper presents the melting behavior and high temperature stability of a novel heat transfer fluid with a
broad operating range that may be suitable for applications in solar thermal power. Future work includes an
economic optimization of the HTF composition as well as full thermophysical property characterization.
Scale up testing will begin with component tests and pilot scale plants before full scale commercial
deployment. The materials discovery methods described in this paper can be used to develop advanced
thermal materials for other heat transfer and thermal storage applications.
This material is based upon work supported by the U.S. Department of Energy under Award Number DE-
 G. J. Kolb, “Conceptual design of an advanced trough utilizing a molten salt working fluid,” presented at
SolarPACES Symposium, Las Vegas, Nevada, 2008.
 “Therminol VP-1 heat transfer fluid by Solutia,” Technical Bulletin 7239115B, Solutia, Inc., 1999.
 “Dowtherm A heat transfer fluid,” Form No. 176-1337-397 AMS, Dow Chemical Company, 1997.
 “Hitec heat transfer salt,” technical bulletin, Coastal Chemical Co., LLC.
 “Hitec solar salt,” technical bulletin, Coastal Chemical Co., LLC.
 C. R. Barkhurst and W. P. Manning, “Hitec heat transfer fluid,” technical whitepaper, Coastal Chemical
Co., LLC, 2004.
 P.L. Lin, A. D. Pelton, and C. W. Bale, “Computation of ternary molten salt phase diagrams,” J.
American Ceramic Soc., vol. 62, no. 7-8, pp. 414-422, 1979.
 G. J. Janz et al., “Physical properties data compilations relevant to energy storage,” NSRDS-NBS 61,
parts I, II, and IV, 1981.
 R. W. Bradshaw and N. P. Siegel, “Development of molten nitrate salt mixtures for concentrating solar
power systems,” paper no. 11538, proc. SolarPACES, Berlin, Germany, 2009.
 Phase Diagrams for Ceramists, American Ceramic Society/NIST, vol. 1-4, 7, 1964-1989.