For the past century, vapor-compression technology has run most of our refrigerators and air conditioners. It’s inexpensive, reliable, and simple. When a gas is sufficiently compressed, it heats up; it then condenses into a liquid, dumping heat to the outside. After the pressure is released, the liquid evaporates and cools the inside surroundings. The changes in entropy produced during those phase transitions are large enough to make the cycle an extremely useful refrigerator.
Unfortunately, its primary coolants, hydrofluorocarbons, pose an environmental hazard. They are powerful greenhouse gases, with a global warming potential thousands of times that of carbon dioxide. (Their ozone-destroying predecessors, chlorofluorocarbons, were even worse—see the article by Anne Douglass, Paul Newman, and Susan Solomon, Physics Today, July 2014, page 42.) And with growing populations in parts of the world, the demand for refrigeration, and in turn hydrofluorocarbons, is expected to triple by 2050.1
To counter that trend, engineers have looked to certain solid-state materials that offer a more eco-friendly alternative. Known as calorics, they undergo a phase transition in response to an external field and cool the surroundings without involving any greenhouse gases. The temperature changes are induced by an electric field, magnetic field, or mechanical stress, for instance—levers that induce what are known as electrocaloric, magnetocaloric, and elastocaloric effects, respectively. (See the article by Ichiro Takeuchi and Karl Sandeman, Physics Today, December 2015, page 48.)
But those alternative methods come with their own problems. In electrocaloric devices, electric fields of 200 MV/m are common. That’s more than 50 times the dielectric strength of air and can be safely produced only across micrometer-sized films. Some magnetocaloric devices similarly require incredibly strong fields, up to 5 T, which are difficult to produce using permanent magnets alone. And elastocaloric devices with a high temperature span operate at stresses of 800 MPa—four times as large as the stress that steel can endure before deforming. What’s more, the payoff of caloric materials is weak: They produce much smaller cooling power than does a hydrofluorocarbon.
Doctoral student Drew Lilley and his adviser, Ravi Prasher, both at California’s Lawrence Berkeley National Laboratory, now propose a new caloric process they call the ionocaloric effect. The work arose from a question that Prasher posed to his group a decade ago: How can we most efficiently change the melting temperature of a solid? “I realized that none of the existing caloric materials would provide a high-enough temperature lift between the hot and cold side of a refrigeration cycle without the expense of such high fields,” he says. “And it struck me that the most efficient approach would be to diffuse ions into a solvent.”
The ionocaloric effect works on the same principle as an old-fashioned ice-cream maker: Adding salt to a material lowers its melting temperature and can trigger a phase change. The researchers chose sodium iodide as the salt and ethylene carbonate as the solvent material. Unlike other calorics, whose phase transitions take place entirely in the solid state and involve changes only in crystal orientation, ionocalorics manifest a solid–liquid phase change.
In the new work, Lilley and Prasher did more than reproduce an effect that’s evident every time you churn out ice cream at a summer picnic. They described how to make it reversible and, in a proof-of-concept device, demonstrated its feasibility in a closed system.2
Entropy generator
Because the whole medium participates in the phase change, the entropy change produced in the carbonate–salt system is 500 J/K·kg. That’s more than 10 times the entropy change found in state-of-the-art magneto-, electro-, and elastocaloric materials and comparable to that of R134a (today’s most common hydrofluorocarbon, whose entropy change is 650 J/K·kg). The ionocaloric system’s temperature change was large as well, 29 °C.
The large entropy change wasn’t accidental. The researchers had established specific criteria for the ideal properties of the refrigerant. The solvent should have a melting point above room temperature and a eutectic point—the binary salt mixture’s melting point—well below room temperature. It should also have a high latent heat of fusion to maximize the heat absorbed per cycle and a high cryoscopic constant (the depression in a solvent’s melting point on dissolving one mole of a substance in 1000 g of it) to ensure that the large temperature change can be achieved using small amounts of electrolyte.
With those criteria in mind, they identified the ethylene carbonate–sodium iodide (EC–NaI) system as the most promising. At 205 J/mL, the carbonate’s latent heat of fusion is not much less than that of water, whose value (330 J/mL) is among the highest of known molecules near room temperature. Ethylene carbonate is a common additive to lithium-ion battery electrolytes.
The cooling cycle, shown in figure 1, starts with the addition of solid NaI to solid EC close to the solvent’s melting temperature of 36 °C. The temperature of the mixture quickly drops to 6 °C and it partially melts. The liquid-solid slurry is then pumped into a second chamber, where the remaining solids melt by absorbing heat from their surroundings.
The ionocaloric cycle spans four stages: (1) A salt is mixed into a solid solvent, which cools the mixture to its now-lower melting point. (2) The slurry is then pumped into a second chamber, where it melts while absorbing heat from its surroundings. (3) It’s then pumped through a desalination circuit that applies a small voltage to separate the salt from the solvent, a process that raises the solvent’s melting temperature. (4) Finally, the diluted liquid recrystallizes on contact with a hot reservoir at the now-higher melting point. It and the concentrated salt can then restart the cycle. (Adapted from ref. 2.)
The ionocaloric cycle spans four stages: (1) A salt is mixed into a solid solvent, which cools the mixture to its now-lower melting point. (2) The slurry is then pumped into a second chamber, where it melts while absorbing heat from its surroundings. (3) It’s then pumped through a desalination circuit that applies a small voltage to separate the salt from the solvent, a process that raises the solvent’s melting temperature. (4) Finally, the diluted liquid recrystallizes on contact with a hot reservoir at the now-higher melting point. It and the concentrated salt can then restart the cycle. (Adapted from ref. 2.)
To desalinate the mixture and restore the ingredients to their original forms, the researchers pump the liquid to a separator stage, where a voltage is applied across two ion-exchange membranes. The technique they use, electrodialysis, takes as little as 0.22 V to manipulate and separate ions from the solution. Through a clever arrangement of the membranes, the electrodialysis purifies—and therefore reheats—the carbonate: A concentrate of sodium cations and iodide anions flows into one compartment while a dilute liquid of EC collects in another.
The EC next flows to a fourth chamber, where it releases its heat and crystallizes. It is then ready to recombine with the NaI, shuttled through a separate tube, to restart the cycle.
Relative Carnot efficiency
In principle, the larger the entropy change, the larger the cooling energy. But the Achilles’ heel of the system is its slow desalination step, which limits the cycling frequency and therefore the cooling power. The commercial membrane (Nafion) has a flow resistance for the electrolytes in EC–NaI that is about 100 times as large as it is for the water-based systems the membrane was designed for.
Even so, the experimental prototype performs nearly 30% as well as a Carnot refrigerator—the theoretically ideal case. “That efficiency puts us on the map commercially,” says Lilley. In a plot of its relative Carnot efficiency versus cooling power density (figure 2), the ionocaloric prototype exhibits a higher efficiency than most other caloric methods.
The efficiency of the ionocaloric system compares well with that of other caloric prototypes. Two different ionocaloric curves (orange and purple) are presented because the performance depends on the temperature difference maintained between the hot and cold sides. Taken at different operating conditions of the cycle, the data reveal different efficiencies and cooling powers depending on whether the system was run quickly (higher power but lower efficiency) or slowly (lower power but higher efficiency). (Adapted from ref. 2.)
The efficiency of the ionocaloric system compares well with that of other caloric prototypes. Two different ionocaloric curves (orange and purple) are presented because the performance depends on the temperature difference maintained between the hot and cold sides. Taken at different operating conditions of the cycle, the data reveal different efficiencies and cooling powers depending on whether the system was run quickly (higher power but lower efficiency) or slowly (lower power but higher efficiency). (Adapted from ref. 2.)
Conventional hydrofluorocarbon refrigerants are not plotted. Vapor–compression technology would have a far greater cooling power density—on the order of 600 W/L, says Lilley, where the reference volume is that of the compressor—compared with about 10 W/L measured for the ionocaloric prototype at a similar efficiency. Nonetheless, he and Prasher hope to address the difference by reducing the membranes’ resistance. The improvement in ion conductivity would increase the power density of their ionocaloric device.
They have yet to test the system’s durability, but it appears to show little fatigue. “You can repeat the freeze–thaw cycle as many times as you’d like,” Lilley says. The ion-exchange membranes themselves were standard, commercial models, and the researchers have yet to develop others better suited for the electrolytes.
Still, Lilley and Prasher remain optimistic that a practical version of the new refrigerator technology is within reach. They have filed a US patent application.
