Magnetic refrigeration is based on the magnetocaloric effect (MCE), an intrinsic property of all magnetic materials that peaks in the vicinity of the magnetic ordering temperature. In the case of a ferromagnetic material, it is the warming as the magnetic moments of the atoms are aligned on the application of a magnetic field, and the cooling when the magnetic moments become randomly oriented on removing the magnetic field. The warming and the cooling of a magnetic material in response to a changing magnetic field is similar to the warming and the cooling of a gaseous medium in response to compression and expansion. Therefore, MR operates by magnetizing/demagnetizing the magnetic material.

Since the refrigerant is a solid (usually spheres or thin sheets), the heat transfer is provided by a fluid, such as water, water containing antifreeze, or an inert gas, depending on the operating temperature. This replaces the compressor and the gas or the low boiling temperature liquid in conventional refrigeration systems. However, the controlling circuits remain essentially the same.

A schematic of the successful laboratory magnetic refrigerator for near room temperature applications, built by the Astronautics Corporation of America in collaboration with the Ames Laboratory, is shown in Figure 1. The magnetic field is provided by a conventional liquid helium-immersed superconducting niobium titanium (NbTi) solenoid mounted in a Dewar container with a warm bore (a cylindrical opening along the central axis that stays at room temperature). The magnet is run in the persistent mode (i.e. the current continues to flow inside the closed superconducting loop in the absence of an external current source). The refrigerant is the pure lanthanide metal, gadolinium (Gd), which orders ferromagnetically at 21° C (294 K). The two regenerator beds are composed of Gd spheres (a total of 1.5 kg in each bed) that are mounted on a carrier assembly, and alternately inserted in the bore of the magnet using an air cylinder drive. The drive movement takes 1 s.

The magnetized bed inside the magnet is warmed due to the magnetocaloric effect (MCE), and the demagnetized bed outside the magnet is cooled due to the reverse MCE. The heat transfer fluid in this MR is water, which is pumped through the beds and heat exchangers as shown in the diagram. The water picks up heat when passing through the magnetized (hot) bed cooling the Gd inside, and then the heat is dissipated as the water passes through the hot heat exchanger. The heat load is picked up by the water in the cold heat exchanger and is dissipated when passing through the demagnetized (cold) bed warming the Gd inside. The flow is accomplished in approximately 2 s, after which the positions of the beds are interchanged (i.e. the previously demagnetized bed is inserted inside the magnet and the previously magnetized bed is on the outside). The flow of the water is rerouted accordingly and the refrigeration cycle is repeated.

The experimental results of the energy efficiency of this MR system are depicted in Figure 2. The Carnot efficiency has been calculated excluding the energy losses in the seals, which in this refrigerator were standard off-the-shelf seals. The use of special low friction seals would greatly reduce these losses. As can be seen, the efficiency of this MR is about the same order of magnitude as for commercial vapor cycle refrigerators (20 to 30% of Carnot), even at the lowest magnetic field change from 0 to 1.5 T. It reaches 50 to 60% when the magnetic field increases to 5 T.

The overall cooling power that has been obtained ranges from 200 W to 600 W, depending on the magnetic field strength and other operating parameters. From the total volume of the two beds packed with the magnetic refrigerant, which is just under 600 ml, the specific cooling power obtained is about 1 W/ml. The observed coefficient of performance (COP), which is defined as the ratio of cooling power to work input, ranges from 2 to 9. This compares favorably with the COP observed for most standard gas compression devices.

The maximum magnetocaloric effect (MCE) in Gd occurs at the temperature where it orders ferromagnetically (294 K or 21° C). When the magnetic field changes from 0 to 1.5 T, its MCE (i.e. the change in its temperature when the magnetic field changes) is 4.5 K (or 4.5° C), and 11 K (or 11° C) when the magnetic field changes from 0 to 5 T. Nonetheless, as shown on Figure 2, the temperature span achieved exceeds the maximum MCE by a factor of 2 to 3 at a competitive efficiency. This is made possible by using an unusual thermodynamic cycle –- active magnetic regeneration (AMR)-- in which the magnetic refrigerant acts both as a refrigerant and a regenerator. Clearly, the performance of this MR would be enhanced by using a magnetic refrigerant material with a larger MCE than Gd as well as by improving the refrigerator’s design.

Intermetallic alloys recently discovered at the Ames Laboratory provide these much improved magnetic refrigerants. They are formed in the ternary system gadolinium-silicon-germanium (Gd-Si-Ge), and their stoichiometry, Gd₅(Si_xGe_1-x)₄, corresponds to the pseudobinary cross-section Gd₅Si₄ - Gd₅Ge₄. The zero magnetic field phase diagram of this system is shown in Figure 3. The alloys with 0.5 < x £  1, i.e. the Si- rich Gd₅Si₄-based solid solution, are normal ferromagnets. The Gd₅Ge₄-based solid solution extends from 0 £ x £  0.2. The two-step magnetic ordering occurs in this region with the upper transition being from a paramagnet (average alignment of magnetic moments close to random) to a ferrimagnet (partial anti-parallel alignment of magnetic moments), and the lower one from a ferrimagnet to a ferromagnet (parallel alignment of magnetic moments). The upper ordering temperature in the alloys with 0 £  x £  0.2 is almost independent of composition, while the lower transition temperature decreases rapidly as x goes from 0.2 to 0.

The intermediate ternary solid solution phase Gd₅(Si_xGe_1-x)₄ extends from 0.24 £  x £  0.5 and has a monoclinically distorted lattice, derived from the parent orthorhombic samarium germanium (Sm₅Ge₄) type structure formed by both Gd₅Si₄ and Gd₅Ge₄. The alloys in this phase region also order magnetically in two steps: on lowering the temperature they initially order ferromagnetically, and at a slightly lower temperature they undergo a second transition from ferromagnet-I to ferromagnet-II. For all of the alloys with x £  0.5 the upper transition is a second order transition and the lower one is a first order phase transition. The bulk of the magnetic entropy is associated with the lower temperature first order phase transition. This brings about a magnetocaloric effect, which can also be expressed in terms of the magnetic entropy change, D S_mag, exceeding that of previously known lanthanide metals and alloys by a factor of 2 to 7, as shown on Figure 4.

The Gd₅(Si₂Ge₂) alloy represents the terminal composition where the magnetic properties change abruptly on further increase of the silicon content (see Figure 3).

The lower ordering temperature for this alloy is 276 K (3° C), and the giant magnetocaloric effect extends to just above 290 K (17° C) for a magnetic field change from 0 to 5 T. This alloy would make an excellent refrigerant material for a refrigerator with the heat rejected into chilled water. However, for use in a magnetic refrigerator with the heat rejection into the ambient environment (i.e. the temperature of the hot sink would be approximately equal to 25-35^oC, or 298-308 K) one needs a material with an ordering temperature greater than 276 K. By alloying Gd₅(Si₂Ge₂) with trivalent gallium (Ga) substituting for the mixture of tetravalent Si+Ge its properties can be selectively modified.

The more metallic nature of Ga most likely affects the distribution of electrons between the valence and conduction bands, and the small alloying addition of Ga (0.333 at.%) retains the monoclinic crystal structure, which is necessary to preserve the first order magnetic phase transition and to maintain the giant magnetocaloric effect. Simultaneously, it probably enhances the exchange interactions which causes the increase of both the upper and the lower ordering temperatures in the Gd₅(Si_1.985Ge_1.985Ga_0.03) alloy. In fact, the lower (first order phase transition) occurs at ~ 286 K (13° C), and the 0 to 5 T magnetic field change extends the region of the giant magnetocaloric effect from ~ 290 (17° C ) to 310 K (37° C), as shown in Figure 4.

Another important parameter characterizing the potential for the use of magnetic refrigerants is the refrigerant capacity, which is defined as:

where T₁ and T₂ are the temperatures of the hot and cold sinks, respectively, and D S_mag (T) is the refrigerant’s magnetic entropy change as a function of temperature. The refrigerant capacity, therefore, is a measure of how much heat can be transferred between the cold and hot sinks in one ideal refrigeration cycle. A comparison, shown in Figure 5, of some of the Gd₅(Si_xGe_1-x)₄ alloys with the best known refrigerants in the corresponding ranges of temperature shows that the new materials have 25 to 120% more capacity than any known magnetic refrigerant. This should translate into a corresponding increase in the MR performance provided that all other key factors, such as the MR design and the heat transfer characteristics, are unchanged.

The Gd₅(Si_xGe_1-x)₄ series of alloys, where 0 £  x £  0.5, plus the Gd₅(Si_1.985Ge_1.985Ga_0.03) alloy have an extremely large magnetocaloric effect exceeding that in known prototypes by as much as a factor of 2 (in terms of refrigerant capacity) and by a factor of 7 (in terms of D S_mag). Another unique feature of this series of materials is that the magnetic ordering temperature where the giant magnetocaloric effect exists can be easily tuned between room temperature and the liquefaction temperature of hydrogen gas. Therefore, the discovery of the large magnetic cooling capacity of the Gd₅(Si_1-xGe_x)₄ materials, combined with the successful design of the laboratory scale magnetic refrigerator, offer the long awaited breakthroughs in active magnetic regenerator magnetic refrigeration technology. It is anticipated that the widespread commercialization of this cooling technology will take place in time to provide large benefits in energy savings, in the reduction of greenhouse gases, and in the preservation of the environment.

At the present stage of magnetic refrigeration technology expanded support from industry and government would make possible a rapid transition from the laboratory scale (proof of the principle) refrigerator to commercially assembled, highly efficient and reliable magnetic refrigerator units. Magnetic refrigeration is unique because it is potentially scaleable in size and power (from Watts to MegaWatts of cooling power near room temperature) without significant losses in efficiency. In contrast, scaling down is intrinsically difficult with conventional gas compression technology. Government and industrial support is necessary to continue studies of the basic phenomena related to the magnetocaloric effect, and the engineering fundamentals of the performance of new laboratory scale magnetic refrigerators powered by either superconducting or permanent magnets. New laboratory scale magnetic refrigerator units employing different designs need to be built and comprehensively tested with known and new magnetic refrigerant materials. This will result in an understanding of the fundamental relations between composition, crystal, and magnetic structures, and the magnetothermal properties of solid magnetic materials; and also the behavior of magnetic materials in an environment of varying magnetic fields. The ultimate goal is the design of better magnetic refrigerants and more energy efficient magnetic refrigerators.

Karl A. Gschneidner, Jr., Ph.D., Ames Laboratory, Iowa State University, Ames, IA 50011-3020.
Phone: (515)-294-7931 Fax: (515)-294-9579
E-mail: CAGEY@ameslab.gov

Vitalij K. Pecharsky, Ph.D., Ames Laboratory, Phone: (515)-294-8220. Fax: (515)-294-9579. E-mail: VITKP@ameslab.gov