Molten salts run concentrated solar plants, carry heat in advanced reactors, and electrolyze the world’s aluminium — and the property data those systems are designed around are far shakier than their users assume. The authoritative reference correlations for the thermal conductivity of common inorganic molten salts carry stated uncertainties reaching 15–20% for several salts at 95% confidence, while the viscosity correlations for the same salts sit near 1–7%12. That gap is not sloppiness; it is what happens when a measurement must be performed inside an incandescent, chemically aggressive, convecting, radiating liquid. This article explains why hot liquids defeat ordinary thermal instruments, which properties actually drive storage and reactor engineering, how integrated melt testers gather density, surface tension, conductivity and crystallization behavior in one architecture, and what to demand of any molten-salt number before designing on it.
| Measured directly | Density, surface tension, electrical conductivity, initial crystallization temperature; α by flash with purpose-built containment |
|---|---|
| Derived | k from α with ρ and cp; storage sizing from the property set |
| Required inputs | Composition as measured, atmosphere and moisture history, direct sample temperature |
| Direction | Isotropic liquid — direction is not the variable; containment and convection are |
| Main corrections | Convection suppression, radiative transfer in semi-transparent melts, crucible compatibility |
| Reported uncertainty | Benchmark against critically assessed reference correlations |
- 1Why hot liquids defeat ordinary thermal instruments
- 2The data gap, quantified
- 3The property set and what each one decides
- 4Interactive: how property uncertainty propagates into a storage design
- 5How integrated melt testers work
- 6What to demand of a molten-salt number
- 7Frequently asked questions
- 8Keep exploring the knowledge hub
- 9References
Why hot liquids defeat ordinary thermal instruments
Four adversaries arrive together. Convection: a liquid heated from anywhere begins to move, and moving liquid transports heat by a mechanism no conduction model accounts for — so measurements must either suppress convection geometrically or restrict themselves to time windows shorter than its onset1. Radiation: many molten salts are semi-transparent in the infrared, so energy travels through the sample rather than diffusing along it, inflating apparent conductivity unless the analysis accounts for it. Together with convection it is a principal reason conductivity reference values carry uncertainties several times those of viscosity, though the relative weight of the two varies with salt, geometry and method13. Chemistry and containment: hot salts attack crucibles, wet or fail to wet surfaces unpredictably, dissolve sensors, evaporate and absorb moisture from the air — containment compatibility deserves the same design attention as the measurement model; nitrates decompose above their working range, carbonates evolve gas, chlorides and fluorides corrode aggressively, so containment materials become part of the measurement45. Temperature fields: a property quoted at 600 °C means little if the sample spans twenty degrees, which is why direct sample-temperature measurement rather than furnace-setpoint inference is a design priority in modern testers.
The data gap, quantified
The scale of the problem is documented rather than anecdotal. The 2018 reference-correlation effort for thermal conductivity examined all available primary data for thirteen inorganic molten salts and could only assign uncertainties of roughly 7% for the best-characterized nitrates, rising to 15–20% for several chlorides and bromides1. Its companion viscosity study a year later — same salts, same critical-assessment methodology — produced correlations at roughly 1–7%2. A recent fluoride-salt assessment repeats the pattern for reactor-relevant compositions, noting how wide the variability among independent measurements of the same compound remains5. Independent measurement campaigns tell the same story from the bench: eutectic carbonate work found the thermal diffusivity route required a special crucible design and a purpose-built sample-preparation method before flash analysis could be applied at all4, and viscosity studies on single salts routinely report that different laboratories’ datasets disagree by more than any of their individual uncertainty statements67. The engineering consequence is blunt: a molten-salt heat exchanger designed on a handbook conductivity is designed on a number that may be a fifth wrong, and the honest response is to measure the actual working composition.
The property set and what each one decides
| Property | What it decides | Measurement difficulty |
|---|---|---|
| Density ρ | Tank volumes and storage capacity; also an explicit input to oscillating-cup viscometry and implicit in calibrated rotational cells8 | Moderate — Archimedean and maximum-bubble routes are mature |
| Surface tension | Wetting and containment behavior, bubble and droplet phenomena, meniscus errors in other instruments8 | Moderate — but sensitive to purity and atmosphere |
| Viscosity η | Pumping power, flow regime and convective heat-transfer coefficients2 | Demanding — oscillating-cup and rotational methods with careful correction3 |
| Thermal conductivity k | Heat-exchanger area and temperature approach | Hardest — convection and radiation both attack it1 |
| Heat capacity cp | Energy stored per kilogram — the economics of thermal storage | Moderate — calorimetry, with purity sensitivity4 |
| Initial crystallization temperature | Freeze-protection margin and minimum operating temperature | Moderate — cooling-curve analysis, hysteresis-aware4 |
| Electrical conductivity | Electrolysis processes and in-situ diagnostics | Moderate — electrode compatibility is the constraint |
Interactive: how property uncertainty propagates into a storage design
The simulator takes the data gap out of the abstract. It sizes a simple thermal-storage duty — energy stored and the heat-exchanger area needed to move it — and propagates the uncertainties you dial on heat capacity and thermal conductivity into bands on both results. Set conductivity to the 15–20% that reference correlations honestly carry for some salts and watch the required-area band widen into the region where engineering margin, not physics, decides the capital cost.
The sizing is a deliberately simplified teaching calculation — a single lumped duty with linear sensitivities — not a design tool; real systems add flow regime, fouling, transients and safety factors. The transferable point is the leverage: property uncertainty becomes hardware cost through arithmetic no design review can escape9.
How integrated melt testers work
The instrument answer to a multi-property problem is a shared high-temperature core with interchangeable measurement modules. The ThermalSure® Molten Salt Physical Property Comprehensive Tester and its general-melt sibling are built on exactly that architecture: a heating module, lifting module, electrical control, water and gas services, and a measurement module, coordinated by dedicated software — with density, surface tension, electrical conductivity and initial crystallization temperature all addressed on one platform per the published product description, and modules swapped rather than instruments changed between parameters. Which parameters a given configuration covers, and over what temperature envelope and wetted-materials list, should be confirmed against current product documentation. Two design choices carry the physics. Modularity keeps the sample in one thermal environment across parameters, so the properties reported belong to the same melt state rather than to four different afternoons. Direct sample-temperature measurement — reading the melt rather than the furnace — addresses the temperature-field problem that quietly widens hot-liquid error budgets10. Viscosity, whose demands differ enough to deserve its own instrument, is covered by the dedicated viscometer line.
What to demand of a molten-salt number
Five requirements make a molten-salt property usable. Composition as measured, not as ordered — eutectic ratios drift with purity and handling, and properties follow4. Atmosphere and moisture history, because hygroscopic salts change with exposure. Temperature of the sample, measured directly, with the range covered. Method and its known systematics — particularly whether radiative transfer was addressed for conductivity and meniscus effects for viscosity13. Uncertainty with coverage, benchmarked honestly against reference correlations rather than quoted from an instrument brochure11. Numbers carrying all five can be designed on; numbers carrying none belong in a literature review, not a specification — the standard our report-reading guide generalizes across thermal metrology.
Frequently asked questions
Why is molten-salt thermal conductivity so uncertain?
Because convection and radiative transfer both masquerade as conduction in a hot, semi-transparent liquid, and separating them demands geometry and analysis that few campaigns implement identically — which is why critically assessed correlations still carry 15–20% for several salts1.
Can standard laboratory instruments handle molten salts?
Only with purpose-built containment: flash analysis of a molten salt required a special crucible design and dedicated sample preparation before it worked at all4. Corrosion, wetting and decomposition make containment part of the method rather than an accessory5.
Do published correlations cover my salt mixture?
Reference correlations exist for a limited set of pure salts and a few technologically important mixtures; most industrial compositions sit outside them, which is precisely the case for measuring the working fluid rather than interpolating25.
Which property should be measured first?
Density is a sensible first step, since storage sizing starts there and oscillating-cup viscometry consumes it directly; then the property your design is most sensitive to — usually conductivity for heat-exchanger-limited systems and viscosity for pumping-limited ones83.
How does purity affect the results?
Strongly and in every direction: moisture, oxide impurities and off-eutectic composition shift crystallization temperature, viscosity and conductivity together. Purity and handling history belong with every reported value4.
Keep Exploring the ACS Thermal Metrology Knowledge Hub
This article is one chapter of the ACS thermal metrology knowledge hub. To keep going:
- Thermal conductivity & diffusivity testing: the pillar guide — methods, samples and a buyer’s framework in one place.
- High-temperature viscometry — melts, molten salts and liquid metals.
- Thermal measurement at extremes — from cryogenic frost to above 1000 °C.
- Laser flash analysis (LFA) explained — the rear-face transient workhorse, and where it fails.
- How to read a thermal test report — separating measurement from decoration.
- ACS thermal testing services — the team that turns these distinctions into data on your specimen.
References
This article discusses molten-salt and melt thermophysical property testing for educational purposes. Uncertainty figures are those stated in the cited critically assessed reference correlations; instrument descriptions summarize published product information and current specifications are maintained on the respective product pages. The storage simulator is a simplified teaching calculation, not a design tool. Molten salts present serious thermal and chemical hazards — consult safety documentation and qualified engineering support before any handling.